ggml-cpu : fix CPU affinity mask being ignored on Android (#26838 )

ggml : require contiguous src for ROLL on CUDA and Metal (#25928 )
ggml_roll only asserts nb[0] == ggml_type_size, so a permuted src is a valid input, but the CUDA and Metal roll kernels index by ne alone and never read the nb strides. A non-contiguous src therefore produced silently wrong results. Neither backend declared a contiguity requirement in supports_op, so the scheduler did not fall back to the CPU implementation, which does handle strides correctly. Add the requirement to both backends, matching the existing GGML_OP_ROPE guard, and add a permuted test_roll case.
2026-08-10 15:03:36 +02:00 · 2026-08-10 15:13:40 +03:00 · 2026-08-10 15:01:44 +03:00 · 2026-08-10 13:32:51 +02:00 · 2026-08-10 13:32:22 +02:00 · 2026-08-10 13:31:09 +02:00
70 changed files with 2549 additions and 835 deletions
@@ -4,6 +4,10 @@ inputs:
  cuda_version:
    description: "CUDA toolkit version"
    required: true
+  cuda_arch:
+    description: "CUDA target architecture"
+    required: false
+    default: "x64"

 runs:
  using: "composite"
@@ -127,3 +131,26 @@ runs:
          echo "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.3\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
          echo "CUDA_PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.3" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
          echo "CUDA_PATH_V13_3=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.3" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
+
+    - name: Install Cuda Toolkit 13.4 for ARM64
+      if: ${{ inputs.cuda_version == '13.4' && inputs.cuda_arch == 'arm64' }}
+      shell: pwsh
+      run: |
+          mkdir -p "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4"
+          choco install unzip -y
+          curl -O "https://packages.nvidia.com/bin-archive/pool/windows-x86_64/5B515474-7E78-11F1-8656-C51E4F4B317F/cccl-windows-x86_64-13.3.4.1.2-archive.zip"
+          curl -O "https://packages.nvidia.com/bin-archive/pool/windows-x86_64/5B515474-7E78-11F1-8656-C51E4F4B317F/cuda_crt-windows-x86_64-13.4.46-archive.zip"
+          curl -O "https://packages.nvidia.com/bin-archive/pool/windows-x86_64/5B515474-7E78-11F1-8656-C51E4F4B317F/cuda_nvcc-windows-x86_64-13.4.46-archive.zip"
+          curl -O "https://packages.nvidia.com/bin-archive/pool/windows-x86_64/5B515474-7E78-11F1-8656-C51E4F4B317F/libnvvm-windows-x86_64-13.4.46-archive.zip"
+          curl -O "https://packages.nvidia.com/bin-archive/pool/windows-arm64/5B515474-7E78-11F1-8656-C51E4F4B317F/cuda_cudart-windows-arm64-13.4.46-archive.zip"
+          curl -O "https://packages.nvidia.com/bin-archive/pool/windows-arm64/5B515474-7E78-11F1-8656-C51E4F4B317F/libcublas-windows-arm64-13.7.0.10-archive.zip"
+          unzip '*.zip' -d "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4"
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4\cccl-windows-x86_64-13.3.4.1.2-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4\cuda_crt-windows-x86_64-13.4.46-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4\cuda_nvcc-windows-x86_64-13.4.46-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4\libnvvm-windows-x86_64-13.4.46-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4\cuda_cudart-windows-arm64-13.4.46-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4\libcublas-windows-arm64-13.7.0.10-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" /E /I /H /Y
+          echo "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+          echo "CUDA_PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
+          echo "CUDA_PATH_V13_4=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.4" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
@@ -848,6 +848,7 @@ jobs:
          name: llama-bin-win-${{ matrix.backend }}-${{ matrix.arch }}.zip

  windows-cuda:
+    name: windows-cuda (${{ matrix.cuda }}, ${{ matrix.arch }})
    needs: [check-release]
    if: ${{ needs.check-release.outputs.should_release == 'true' }}

@@ -858,7 +859,16 @@ jobs:

    strategy:
      matrix:
-        cuda: ['12.4', '13.3']
+        include:
+          - cuda: '12.4'
+            arch: x64
+            defines: '-DGGML_CUDA_CUB_3DOT2=ON'
+          - cuda: '13.3'
+            arch: x64
+            defines: ''
+          - cuda: '13.4'
+            arch: arm64
+            defines: '-DCMAKE_TOOLCHAIN_FILE=cmake/arm64-windows-msvc-cuda.cmake'

    steps:
      - name: Clone
@@ -876,6 +886,7 @@ jobs:
        uses: ./.github/actions/windows-setup-cuda
        with:
          cuda_version: ${{ matrix.cuda }}
+          cuda_arch: ${{ matrix.arch }}

      - name: Install Ninja
        id: install_ninja
@@ -885,54 +896,62 @@ jobs:
      - name: ccache
        uses: ggml-org/ccache-action@v1.2.21
        with:
-          key: release-windows-2022-x64-cuda-${{ matrix.cuda }}
+          key: release-windows-2022-${{ matrix.arch }}-cuda-${{ matrix.cuda }}

      - name: Build
        id: cmake_build
        shell: cmd
        # TODO: Remove GGML_CUDA_CUB_3DOT2 flag once CCCL 3.2 is bundled within CTK and that CTK version is used in this project
        run: |
-          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" x64
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" ${{ matrix.arch == 'x64' && 'x64' || 'amd64_arm64' }}
          cmake -S . -B build -G "Ninja Multi-Config" ^
            -DGGML_BACKEND_DL=ON ^
            -DGGML_NATIVE=OFF ^
            -DGGML_CPU=OFF ^
            -DGGML_CUDA=ON ^
-            -DLLAMA_BUILD_BORINGSSL=ON ^
-            -DGGML_CUDA_CUB_3DOT2=ON
+            -DLLAMA_BUILD_BORINGSSL=ON ${{ matrix.defines }}
          set /A NINJA_JOBS=%NUMBER_OF_PROCESSORS%-1
          cmake --build build --config Release -j %NINJA_JOBS% --target ggml-cuda

      - name: ccache-clear
        uses: ./.github/actions/ccache-clear
        with:
-          key: release-windows-2022-x64-cuda-${{ matrix.cuda }}
+          key: release-windows-2022-${{ matrix.arch }}-cuda-${{ matrix.cuda }}

      - name: Pack artifacts
        id: pack_artifacts
        run: |
-          7z a -snl llama-bin-win-cuda-${{ matrix.cuda }}-x64.zip .\build\bin\Release\ggml-cuda.dll
+          7z a -snl llama-bin-win-cuda-${{ matrix.cuda }}-${{ matrix.arch }}.zip .\build\bin\Release\ggml-cuda.dll

      - name: Upload artifacts
        uses: actions/upload-artifact@v6
        with:
-          path: llama-bin-win-cuda-${{ matrix.cuda }}-x64.zip
-          name: llama-bin-win-cuda-${{ matrix.cuda }}-x64.zip
+          path: llama-bin-win-cuda-${{ matrix.cuda }}-${{ matrix.arch }}.zip
+          name: llama-bin-win-cuda-${{ matrix.cuda }}-${{ matrix.arch }}.zip

-      - name: Copy and pack Cuda runtime
+      - name: Copy and pack Cuda runtime (x64)
+        if: ${{ matrix.arch == 'x64' }}
        run: |
          echo "Cuda install location: ${{ env.CUDA_PATH }}"
          $dst='.\build\bin\cudart\'
          robocopy "${{env.CUDA_PATH}}\bin" $dst cudart64_*.dll cublas64_*.dll cublasLt64_*.dll
          robocopy "${{env.CUDA_PATH}}\lib" $dst cudart64_*.dll cublas64_*.dll cublasLt64_*.dll
          robocopy "${{env.CUDA_PATH}}\bin\x64" $dst cudart64_*.dll cublas64_*.dll cublasLt64_*.dll
-          7z a cudart-llama-bin-win-cuda-${{ matrix.cuda }}-x64.zip $dst\*
+          7z a cudart-llama-bin-win-cuda-${{ matrix.cuda }}-${{ matrix.arch }}.zip $dst\*
+
+      - name: Copy and pack Cuda runtime (ARM64)
+        if: ${{ matrix.arch == 'arm64' }}
+        run: |
+          echo "Cuda install location: ${{ env.CUDA_PATH }}"
+          $dst='.\build\bin\cudart\'
+          robocopy "${{env.CUDA_PATH}}\bin\arm64" $dst cudart64_*.dll cublas64_*.dll cublasLt64_*.dll
+          7z a cudart-llama-bin-win-cuda-${{ matrix.cuda }}-${{ matrix.arch }}.zip $dst\*

      - name: Upload Cuda runtime
        uses: actions/upload-artifact@v6
        with:
-          path: cudart-llama-bin-win-cuda-${{ matrix.cuda }}-x64.zip
-          name: cudart-llama-bin-win-cuda-${{ matrix.cuda }}-x64.zip
+          path: cudart-llama-bin-win-cuda-${{ matrix.cuda }}-${{ matrix.arch }}.zip
+          name: cudart-llama-bin-win-cuda-${{ matrix.cuda }}-${{ matrix.arch }}.zip

  windows-sycl:
    needs: [check-release]
@@ -1679,6 +1698,7 @@ jobs:
            - [Windows arm64 (OpenCL Adreno)](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/llama-${{ steps.tag.outputs.name }}-bin-win-opencl-adreno-arm64.zip)
            - [Windows x64 (CUDA 12)](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/llama-${{ steps.tag.outputs.name }}-bin-win-cuda-12.4-x64.zip) - [CUDA 12.4 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/cudart-llama-bin-win-cuda-12.4-x64.zip)
            - [Windows x64 (CUDA 13)](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/llama-${{ steps.tag.outputs.name }}-bin-win-cuda-13.3-x64.zip) - [CUDA 13.3 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/cudart-llama-bin-win-cuda-13.3-x64.zip)
+            - [Windows arm64 (CUDA 13) (preview)](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/llama-${{ steps.tag.outputs.name }}-bin-win-cuda-13.4-arm64.zip) - [CUDA 13.4 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/cudart-llama-bin-win-cuda-13.4-arm64.zip)
            - [Windows x64 (Vulkan)](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/llama-${{ steps.tag.outputs.name }}-bin-win-vulkan-x64.zip)
            - [Windows x64 (OpenVINO)](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/llama-${{ steps.tag.outputs.name }}-bin-win-openvino-${{ needs.windows-openvino.outputs.openvino_version }}-x64.zip)
            - [Windows x64 (SYCL)](https://github.com/ggml-org/llama.cpp/releases/download/${{ steps.tag.outputs.name }}/llama-${{ steps.tag.outputs.name }}-bin-win-sycl-x64.zip)
@@ -25,6 +25,12 @@ on:
      'tools/server/**.*'
    ]

+  pull_request:
+    types: [opened, synchronize, reopened]
+    paths: [
+      '.github/workflows/server-sanitize.yml'
+    ]
+
 env:
  LLAMA_ARG_LOG_COLORS: 1
  LLAMA_ARG_LOG_PREFIX: 1
@@ -90,15 +96,18 @@ jobs:

      - name: Python setup
        id: setup_python
-        uses: actions/setup-python@v6
-        with:
-          python-version: '3.11'
-          pip-install: -r tools/server/tests/requirements.txt
+        uses: actions/setup-python@v7
+
+      - name: Install Python dependencies
+        run: |
+          python3 -m venv .venv
+          .venv/bin/pip install -r tools/server/tests/requirements.txt

      - name: Tests
        id: server_integration_tests
        if: ${{ (!matrix.disabled_on_pr || !github.event.pull_request) }}
        run: |
+          source .venv/bin/activate
          cd tools/server/tests
          export ${{ matrix.extra_args }}
          pytest -v -x -m "not slow"
@@ -107,6 +116,7 @@ jobs:
        id: server_integration_tests_slow
        if: ${{ (github.event.schedule || github.event.inputs.slow_tests == 'true') && matrix.build_type == 'Release' }}
        run: |
+          source .venv/bin/activate
          cd tools/server/tests
          export ${{ matrix.extra_args }}
          SLOW_TESTS=1 pytest -v -x
@@ -0,0 +1,26 @@
+# Used to cross-compile ggml-cuda for Windows ARM64 on an x64 Windows host.
+set( CMAKE_SYSTEM_NAME Windows )
+set( CMAKE_SYSTEM_PROCESSOR arm64 )
+
+if ( DEFINED CUDAToolkit_ROOT )
+    file( TO_CMAKE_PATH "${CUDAToolkit_ROOT}" CUDA_ROOT )
+elseif ( DEFINED ENV{CUDA_PATH} )
+    file( TO_CMAKE_PATH "$ENV{CUDA_PATH}" CUDA_ROOT )
+else()
+    message( FATAL_ERROR "Set CUDAToolkit_ROOT or CUDA_PATH to a Windows CUDA Toolkit with ARM64 target libraries" )
+endif()
+
+if ( DEFINED ENV{VCToolsInstallDir} )
+    file( TO_CMAKE_PATH "$ENV{VCToolsInstallDir}" MSVC_TOOLS_ROOT )
+    set( CMAKE_CUDA_HOST_COMPILER "${MSVC_TOOLS_ROOT}/bin/Hostx64/arm64/cl.exe" CACHE FILEPATH "" )
+endif()
+
+set( CMAKE_CUDA_COMPILER "${CUDA_ROOT}/bin/nvcc.exe" CACHE FILEPATH "" )
+set( CMAKE_CUDA_FLAGS_INIT "-target-dir=arm64" )
+
+# FindCUDAToolkit selects lib/x64 from the host architecture on Windows.
+set( CUDA_CUDART              "${CUDA_ROOT}/lib/arm64/cudart.lib"   CACHE FILEPATH "" )
+set( CUDA_cudart_LIBRARY      "${CUDA_ROOT}/lib/arm64/cudart.lib"   CACHE FILEPATH "" )
+set( CUDA_cublas_LIBRARY      "${CUDA_ROOT}/lib/arm64/cublas.lib"   CACHE FILEPATH "" )
+set( CUDA_cublasLt_LIBRARY    "${CUDA_ROOT}/lib/arm64/cublasLt.lib" CACHE FILEPATH "" )
+set( CUDA_cuda_driver_LIBRARY "${CUDA_ROOT}/lib/arm64/cuda.lib"     CACHE FILEPATH "" )
@@ -3312,8 +3312,9 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
        {"--tools-runtime"}, "OPTION",
        "experimental: run tools in a separate runtime environment (default: none, use host environment)\n"
        "available options:\n"
-        "  'docker:<image>': spin up a new Docker container and reuse it for all invocations, clean up on server exit\n"
-        "  'docker-container:<id>': use an existing Docker container by ID, won't stop on server exit\n",
+        "  'docker:<image>', 'podman:<image>': spin up a new container and reuse it for all invocations, clean up on server exit\n"
+        "  'docker-container:<id>', 'podman-container:<id>': use an existing container by ID, won't stop on server exit\n"
+        "  'ssh:<target>': run tools on a remote POSIX host over SSH, key-based auth and a trusted host key are required\n",
        [](common_params & params, const std::string & value) {
            params.server_tools_runtime = value;
        }
@@ -3086,6 +3086,151 @@ static common_chat_params common_chat_params_init_minicpm5(const common_chat_tem
    return data;
 }

+// An assistant turn is rendered as one or more messages, each
+// "<|start|>assistant to=<recipient><|message|>{content}{END}" where END is
+// <|eom|> (more messages follow) or <|eot|> (end of turn):
+//   - chain-of-thought: to=self, terminated by <|eom|>
+//   - final answer:     to=user, terminated by <|eot|>
+// The generation prompt is just "<|start|>assistant"; the model emits its own
+// " to=...<|message|>".
+static common_chat_params common_chat_params_init_muse_glimmer(const common_chat_template &          tmpl,
+                                                               const autoparser::generation_params & inputs) {
+    common_chat_params data;
+
+    data.prompt            = common_chat_template_direct_apply_impl(tmpl, inputs);
+    data.generation_prompt = "<|start|>assistant";
+    data.format            = COMMON_CHAT_FORMAT_PEG_NATIVE;
+    data.supports_thinking = true;
+
+    data.preserved_tokens = {
+        "<|start|>", "<|message|>", "<|eom|>", "<|eot|>",
+        // ATEM tool-call markup emitted on " to=<tool>" turns.
+        "<atem:function_calls>", "<atem:invoke", "<atem:parameter", "</atem:parameter>",
+        "</atem:invoke>", "</atem:function_calls>",
+    };
+
+    data.message_delimiters = {
+        { COMMON_CHAT_ROLE_ASSISTANT, "<|start|>assistant" },
+        { COMMON_CHAT_ROLE_USER,      "<|start|>user"      },
+        { COMMON_CHAT_ROLE_SYSTEM,    "<|start|>system"    },
+        { COMMON_CHAT_ROLE_TOOL,      "<|start|>tool"      },
+    };
+
+    if (inputs.has_continuation()) {
+        const auto & msg = inputs.continue_msg;
+
+        data.generation_prompt = "<|start|>assistant to=self<|message|>" + msg.reasoning_content;
+        if (inputs.continue_final_message == COMMON_CHAT_CONTINUATION_CONTENT) {
+            data.generation_prompt += "<|eom|><|start|>assistant to=user<|message|>" + msg.render_content();
+        }
+
+        data.prompt += data.generation_prompt;
+    }
+
+    auto extract_reasoning = inputs.reasoning_format != COMMON_REASONING_FORMAT_NONE;
+
+    auto has_tools = inputs.tools.is_array() && !inputs.tools.empty();
+    // Constrained grammar whenever tools are offered.
+    auto include_grammar = has_tools && inputs.tool_choice != COMMON_CHAT_TOOL_CHOICE_NONE;
+
+    auto parser = build_chat_peg_parser([&](common_chat_peg_builder & p) {
+        auto start = p.rule("start", p.literal("<|start|>assistant"));
+
+        if (!extract_reasoning && !include_grammar) {
+            return start + p.content(p.rest());
+        }
+
+        if (extract_reasoning) {
+            p.rule("analysis", p.literal(" to=self<|message|>") + p.reasoning(p.until("<|eom|>")) + p.literal("<|eom|>"));
+        } else {
+            p.rule("analysis", p.literal(" to=self<|message|>") + p.content(p.until("<|eom|>")) + p.literal("<|eom|>"));
+        }
+        auto analysis = p.ref("analysis");
+
+        auto recipient  = p.optional(p.literal(" to=user"));
+        auto final_msg  = p.rule("final", recipient + p.literal("<|message|>") + p.content(p.until("<|eot|>")));
+
+        if (has_tools && inputs.tool_choice != COMMON_CHAT_TOOL_CHOICE_NONE) {
+            auto string_value = p.ac(
+                p.tool_arg_string_value(p.until("</atem:parameter>")) + p.tool_arg_close(p.literal("</atem:parameter>")),
+                "</atem:parameter>");
+
+            auto tool_choice = p.choice();
+            foreach_function(inputs.tools, [&](const json & tool) {
+                const auto &      function = tool.at("function");
+                const std::string name     = function.at("name");
+                auto              params   = function.contains("parameters") ? function.at("parameters") : json::object();
+
+                auto args = p.eps();
+                if (params.contains("properties") && params.at("properties").is_object() && !params.at("properties").empty()) {
+                    auto schema_info = common_schema_info();
+                    schema_info.resolve_refs(params);
+
+                    auto arg_choice = p.choice();
+                    for (const auto & [prop_name, prop_schema] : params.at("properties").items()) {
+                        auto value_parser = p.eps();
+                        if (schema_info.resolves_to_string(prop_schema)) {
+                            value_parser = string_value;
+                        } else {
+                            value_parser = p.tool_arg_json_value(
+                                    p.schema(p.json(), "tool-" + name + "-arg-" + prop_name + "-schema", prop_schema, false))
+                                + p.tool_arg_close(p.literal("</atem:parameter>"));
+                        }
+
+                        auto arg_rule = p.tool_arg(
+                            p.tool_arg_open(p.literal("<atem:parameter name=\"") + p.tool_arg_name(p.literal(prop_name)) + p.literal("\">")) +
+                            value_parser);
+
+                        arg_choice |= arg_rule;
+                    }
+                    args = p.zero_or_more(arg_choice + p.space());
+                }
+
+                auto tool_parser = p.tool(
+                    p.tool_open(p.literal(" to=") + p.until("<|message|>") +
+                                p.literal("<|message|><atem:function_calls>") + p.space() +
+                                p.literal("<atem:invoke name=\"") + p.tool_name(p.literal(name)) + p.literal("\">") + p.space())
+                    << p.tool_args(args)
+                    << p.tool_close(p.literal("</atem:invoke>") + p.space() + p.literal("</atem:function_calls>")));
+
+                tool_choice |= p.rule("tool-" + name, tool_parser);
+            });
+
+            auto tool_calls = inputs.parallel_tool_calls
+                ? p.trigger_rule("tool-call", tool_choice + p.zero_or_more(p.literal("<|eom|>") + start + tool_choice))
+                : p.trigger_rule("tool-call", tool_choice);
+
+
+            if (inputs.tool_choice == COMMON_CHAT_TOOL_CHOICE_REQUIRED) {
+                return p.zero_or_more(start + analysis) + start + tool_calls;
+            }
+            return p.zero_or_more(start + analysis) + start + (tool_calls | final_msg);
+        }
+
+        return p.zero_or_more(start + analysis) + start + final_msg;
+    });
+
+    data.parser = parser.save();
+
+    if (include_grammar) {
+        data.grammar_lazy = inputs.tool_choice != COMMON_CHAT_TOOL_CHOICE_REQUIRED;
+        data.grammar      = build_grammar([&](const common_grammar_builder & builder) {
+            foreach_function(inputs.tools, [&](const json & tool) {
+                const auto & function = tool.at("function");
+                auto         schema   = function.contains("parameters") ? function.at("parameters") : json::object();
+                builder.resolve_refs(schema);
+            });
+            parser.build_grammar(builder, data.grammar_lazy);
+        });
+        data.grammar_triggers = {
+            { COMMON_GRAMMAR_TRIGGER_TYPE_PATTERN,
+              "<\\|start\\|>assistant( to=(?!self<\\|message\\|>)(?!user<\\|message\\|>)[^<]*?<\\|message\\|>)" },
+        };
+    }
+
+    return data;
+}
+
 static json common_chat_extra_context() {
    json ctx = json::object();
    std::chrono::system_clock::time_point now = std::chrono::system_clock::now();
@@ -3114,6 +3259,12 @@ std::optional<common_chat_params> common_chat_try_specialized_template(
        return common_chat_params_init_gpt_oss(tmpl, params);
    }

+    // Muse Glimmer format using " to=<recipient>" recipients and <|eom|>/<|eot|> message terminators.
+    if (src.find("<atem:function_calls>") != std::string::npos && src.find("<|eom|>") != std::string::npos) {
+        LOG_DBG("Using specialized template: Muse Glimmer\n");
+        return common_chat_params_init_muse_glimmer(tmpl, params);
+    }
+
    // Functionary v3.2 - uses recipient-based format with >>>recipient\n{content}
    // Detection: template has ">>>all" for content and ">>>" prefix for tool calls
    if (src.find(">>>all") != std::string::npos && src.find(">>>${recipient}") != std::string::npos) {
@@ -1032,7 +1032,14 @@ struct common_speculative_impl_draft_dflash : public common_speculative_impl {
            return true;
        }

-        if (batch_in.token == nullptr || batch_in.embd != nullptr) {
+        // Target prefill may contain token IDs or multimodal embeddings. Both
+        // produce the target-layer features used to seed the draft KV cache, so
+        // skipping the embedding batches leaves a hole in the draft's cache and
+        // the next injection fails to initialize.
+        // TODO: revisit after https://github.com/ggml-org/llama.cpp/pull/24669 is merged
+        const bool has_tokens     = batch_in.token != nullptr;
+        const bool has_embeddings = batch_in.embd  != nullptr;
+        if (has_tokens == has_embeddings) {
            return true;
        }

@@ -183,6 +183,8 @@ TEXT_MODEL_MAP: dict[str, str] = {
    "Olmo3ForCausalLM": "olmo",
    "OlmoForCausalLM": "olmo",
    "OlmoeForCausalLM": "olmo",
+    "MuseGlimmerAssistantModel": "muse_glimmer",
+    "MuseGlimmerForConditionalGeneration": "muse_glimmer",
    "OpenELMForCausalLM": "openelm",
    "OrionForCausalLM": "orion",
    "PLMForCausalLM": "plm",
@@ -298,6 +300,7 @@ MMPROJ_MODEL_MAP: dict[str, str] = {
    "MiniCPMV4_6ForConditionalGeneration": "minicpm",
    "Mistral3ForConditionalGeneration": "llava",
    "NemotronH_Nano_VL_V2": "nemotron",
+    "MuseGlimmerForConditionalGeneration": "muse_glimmer",
    "PaddleOCRVisionModel": "ernie",
    "Phi4ForCausalLMV": "phi",
    "Qwen2AudioForConditionalGeneration": "ultravox",
@@ -0,0 +1,179 @@
+from __future__ import annotations
+
+import json
+from typing import Any, Iterable, TYPE_CHECKING
+
+import torch
+
+if TYPE_CHECKING:
+    from torch import Tensor
+
+from .base import MmprojModel, ModelBase, TextModel, gguf
+
+
+def _unpermute_for_rope(tensor: "Tensor", n_heads: int) -> "Tensor":
+    """Invert transformers' `_permute_for_rope`: HF stores Q/K in rotate_half layout,
+    llama.cpp consumes the interleaved (NORM) layout."""
+    if tensor.ndim == 2:
+        dim1, dim2 = tensor.shape
+        return tensor.view(n_heads, 2, dim1 // n_heads // 2, dim2).transpose(1, 2).reshape(dim1, dim2)
+    if tensor.ndim == 1:
+        (dim1,) = tensor.shape
+        return tensor.view(n_heads, 2, dim1 // n_heads // 2).transpose(1, 2).reshape(dim1)
+    raise ValueError(f"_unpermute_for_rope: unexpected shape {tuple(tensor.shape)}")
+
+
+@ModelBase.register("MuseGlimmerForConditionalGeneration")
+class MuseGlimmerModel(TextModel):
+    model_arch = gguf.MODEL_ARCH.MUSE_GLIMMER
+
+    def norm_shift(self, name: str) -> float:
+        # All four layer norms use 1, the final norm uses 0.
+        return 1.0 if name.endswith("layernorm.weight") else 0.0
+
+    def set_vocab(self):
+        self._set_vocab_gpt2()
+
+        from transformers import AutoTokenizer
+        tok = AutoTokenizer.from_pretrained(self.dir_model)
+        eot_id = tok.convert_tokens_to_ids("<|eot|>")
+        if isinstance(eot_id, int) and eot_id >= 0:
+            self.gguf_writer.add_eot_token_id(eot_id)
+
+    def set_gguf_parameters(self):
+        super().set_gguf_parameters()
+        hparams = self.hparams
+
+        self.gguf_writer.add_final_logit_softcapping(hparams["final_logit_softcapping"])
+        self.gguf_writer.add_logit_scale(hparams["output_multiplier"])
+        self.gguf_writer.add_sliding_window(hparams["sliding_window"])
+        self.gguf_writer.add_sliding_window_pattern([t == "sliding_attention" for t in hparams["layer_types"]])
+
+    def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
+        shift = self.norm_shift(name)
+        if shift != 0.0:
+            data_torch = data_torch + shift
+
+        # Invert transformers' `_permute_for_rope` on Q/K, we keep ggml's NORM (interleaved) rope
+        if ".self_attn.q_proj." in name:
+            data_torch = _unpermute_for_rope(data_torch, int(self.hparams["num_attention_heads"]))
+        elif ".self_attn.k_proj." in name:
+            data_torch = _unpermute_for_rope(data_torch, int(self.hparams["num_key_value_heads"]))
+
+        # Synthesize QK-norm weights to absorb qk_scale_factor.
+        # MuseGlimmer implementation: scaleless RMSNorm followed by qk_scale_factor..
+        if bid is not None and name.endswith(f"model.layers.{bid}.self_attn.q_proj.weight"):
+            head_dim = self.hparams["head_dim"]
+            q_scale = float(self.hparams["qk_scale_factor"])
+            yield (
+                self.map_tensor_name(f"model.layers.{bid}.self_attn.q_norm.weight"),
+                torch.full((head_dim,), q_scale, dtype=torch.float32),
+            )
+            yield (
+                self.map_tensor_name(f"model.layers.{bid}.self_attn.k_norm.weight"),
+                torch.ones((head_dim,), dtype=torch.float32),
+            )
+
+        yield from super().modify_tensors(data_torch, name, bid)
+
+
+@ModelBase.register("MuseGlimmerForConditionalGeneration")
+class MuseGlimmerVisionModel(MmprojModel):
+    def get_vision_config(self) -> dict[str, Any] | None:
+        c = self.global_config.get("vision_config")
+        if not c:
+            return None
+        # MuseGlimmer actually uses dynamic size, initialize with nominal size
+        image_size = c["pos_emb_height"] * c["patch_size"] * c["merge_size"]
+        return {**c, "image_size": image_size}
+
+    def set_gguf_parameters(self):
+        super().set_gguf_parameters()
+        assert self.hparams_vision is not None
+        c = self.hparams_vision  # enriched vision_config from get_vision_config()
+
+        self.gguf_writer.add_clip_projector_type(gguf.VisionProjectorType.MUSE_GLIMMER)
+        self.gguf_writer.add_vision_attention_layernorm_eps(float(c["layer_norm_eps"]))
+        self.gguf_writer.add_vision_spatial_merge_size(int(c["merge_size"]))
+
+    @classmethod
+    def filter_tensors(cls, item):
+        name, gen = item
+        keep = ("model.vision_tower.", "model.vision_adapter.", "model.vision_projection.")
+        if not any(name.startswith(k) for k in keep):
+            return None
+        return super().filter_tensors((name, gen))
+
+    # 3-layer projector MLP
+    _MM_MLP_MAP = {
+        "model.vision_adapter.fc1": (gguf.MODEL_TENSOR.V_MMPROJ, 0),
+        "model.vision_adapter.fc2": (gguf.MODEL_TENSOR.V_MMPROJ, 1),
+        "model.vision_projection":  (gguf.MODEL_TENSOR.V_MMPROJ, 2),
+    }
+
+    def modify_tensors(self, data_torch, name, bid):
+        assert self.hparams_vision is not None
+        if ".attn.q_proj." in name or ".attn.k_proj." in name:
+            n_heads = int(self.hparams_vision["num_attention_heads"])
+            data_torch = _unpermute_for_rope(data_torch, n_heads)
+        # Lay out the pt=2 temporal slabs of the patch embedding as a conv2d for build_inp()
+        if name.endswith("patch_embedder.patch_embedding.weight"):
+            n_embd = data_torch.shape[0]
+            pt = int(self.hparams_vision["patch_temporal"])
+            ps = int(self.hparams_vision["patch_size"])
+            data_torch = data_torch.view(n_embd, pt, 3, ps, ps).sum(dim=1)  # (n_embd, 3, ps, ps)
+        stem, _, suffix = name.rpartition(".")
+        if stem in self._MM_MLP_MAP:
+            tensor_key, idx = self._MM_MLP_MAP[stem]
+            yield (self.format_tensor_name(tensor_key, bid=idx, suffix="." + suffix), data_torch)
+            return
+        yield (self.map_tensor_name(name), data_torch)
+
+
+@ModelBase.register("MuseGlimmerAssistantModel")
+class MuseGlimmerAssistantModel(TextModel):
+    model_arch = gguf.MODEL_ARCH.DFLASH
+
+    def set_vocab(self):
+        if self.target_model_dir is None:
+            raise ValueError(
+                "MuseGlimmerAssistant (DFlash drafter) requires --target-model-dir pointing to the "
+                "target MuseGlimmer HF directory"
+            )
+
+        original_dir = self.dir_model
+        self.dir_model = self.target_model_dir
+
+        from . import get_model_class
+        with open(self.target_model_dir / "config.json", "r", encoding="utf-8") as f:
+            target_arch = json.load(f)["architectures"][0]
+        target_cls = get_model_class(target_arch)
+        if target_cls is not type(self):
+            target_cls.set_vocab(self)  # ty: ignore[unresolved-attribute]
+        else:
+            super().set_vocab()
+
+        self.dir_model = original_dir
+
+        mask_token_id = self.hparams.get("mask_token_id")
+        if mask_token_id is not None:
+            self.gguf_writer.add_mask_token_id(int(mask_token_id))
+
+    def set_gguf_parameters(self):
+        super().set_gguf_parameters()
+        h = self.hparams
+
+        self.gguf_writer.add_block_size(int(h["block_size"]))
+
+        # dflash.target_layers[k] refers to the inputs going into the ith layer, which come from the (i-1)th layer's output.
+        # The transformers configuration refers to the outputs being recorded.
+        self.gguf_writer.add_target_layers([int(x) + 1 for x in h["target_layer_ids"]])
+
+        if h.get("sliding_window") and h.get("layer_types"):
+            self.gguf_writer.add_sliding_window(int(h["sliding_window"]))
+            self.gguf_writer.add_sliding_window_pattern([t == "sliding_attention" for t in h["layer_types"]])
+
+    def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
+        # DFlash defaults to NEOX (rotate_half) rope, matching transformers HF layout for Q/K, QK-norms
+        # no permutation needed.
+        yield (self.map_tensor_name(name), data_torch)
@@ -197,6 +197,7 @@ class NemotronHModel(GraniteHybridModel):
    """Hybrid mamba2/attention model from NVIDIA"""
    model_arch = gguf.MODEL_ARCH.NEMOTRON_H
    is_moe: bool = False
+    supports_mtp_export = True

    def __init__(self, *args, **kwargs):
        # We have to determine the correct model architecture (MoE vs non-MoE) before
@@ -236,6 +237,25 @@ class NemotronHModel(GraniteHybridModel):
            self._ssm_layers = [i for i, val in enumerate(pattern) if val == "mamba"]
            self._mlp_layers = [i for i, val in enumerate(pattern) if val == "moe"]

+        # `--no-mtp` drops it entirely; `--mtp` exports only the MTP head
+        self._mtp_bid: int | None = None
+        if self.is_moe and not self.no_mtp:
+            n_nextn = self.hparams.get("num_nextn_predict_layers", 0) or 0
+            if n_nextn > 0:
+                assert n_nextn == 1, (
+                    "NemotronH MTP conversion currently supports num_nextn_predict_layers == 1"
+                )
+                self._mtp_bid = self.block_count
+                self.block_count += 1
+                # The folded MTP block carries both an attention sub-layer and a
+                # MoE sub-layer, so register it as both so the per-layer metadata arrays cover it
+                self._attn_layers.append(self._mtp_bid)
+                self._mlp_layers.append(self._mtp_bid)
+                self.tensor_map = gguf.get_tensor_name_map(self.model_arch, self.block_count)
+
+        if self.mtp_only and self._mtp_bid is None:
+            raise ValueError("--mtp was requested, but this model does not contain a supported MTP head")
+
    def get_attn_layers(self):
        pattern = self.hparams.get("hybrid_override_pattern") or self.hparams.get("layers_block_type")
        if pattern is None:
@@ -246,6 +266,36 @@ class NemotronHModel(GraniteHybridModel):

        return [i for i, val in enumerate(pattern) if val == "attention"]

+    @classmethod
+    def filter_tensors(cls, item: tuple[str, Callable[[], Tensor]]) -> tuple[str, Callable[[], Tensor]] | None:
+        name, gen = item
+        if name.startswith("mtp."):
+            # --no-mtp: drop the MTP head entirely
+            if cls.no_mtp:
+                return None
+        elif cls.mtp_only:
+            # --mtp: export the MTP head plus the tensors it shares with the target model
+            keep = name in (
+                "backbone.embeddings.weight",
+                "backbone.norm_f.weight",
+                "lm_head.weight",
+            )
+            if not keep:
+                return None
+        return super().filter_tensors((name, gen))
+
+    def prepare_metadata(self, vocab_only: bool):
+        from_dir = self.fname_out.is_dir()
+        super().prepare_metadata(vocab_only=vocab_only)
+
+        if not self.mtp_only or not from_dir:
+            return
+        output_type: str = self.ftype.name.partition("_")[2]
+        fname_default: str = gguf.naming_convention(
+            self.metadata.name, self.metadata.basename, self.metadata.finetune,
+            self.metadata.version, size_label=None, output_type=output_type, model_type=None)
+        self.fname_out = self.fname_out.parent / f"mtp-{fname_default}.gguf"
+
    def set_gguf_parameters(self):
        super().set_gguf_parameters()

@@ -284,6 +334,10 @@ class NemotronHModel(GraniteHybridModel):
            if (latent_size := self.hparams.get("moe_latent_size")) is not None:
                self.gguf_writer.add_moe_latent_size(latent_size)

+        # MTP head: number of trailing NextN blocks
+        if self._mtp_bid is not None:
+            self.gguf_writer.add_nextn_predict_layers(self.hparams["num_nextn_predict_layers"])
+
    def set_vocab(self):
        # The NemotronH config uses pattern characters (e.g. '-') that may not
        # be supported by the installed transformers version. AutoTokenizer
@@ -350,15 +404,24 @@ class NemotronHModel(GraniteHybridModel):
        if not self.is_moe:
            self.gguf_writer.add_add_bos_token(True)

-    def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
-        if self.is_moe and bid is not None:
-            # Skip Multi-Token Prediction (MTP) tensors. These are used for
-            # for speculative decoding but we don't include them in this model
-            # conversion. See https://github.com/ggml-org/llama.cpp/pull/18886
-            if name.startswith("mtp."):
-                logger.info(f"gguf: Skipping MTP (Speculative) layer: {name}")
-                return
+    _MTP_SPECIAL_RENAMES = {
+        "mtp.layers.0.enorm.weight":           "model.layers.{bid}.enorm.weight",
+        "mtp.layers.0.hnorm.weight":           "model.layers.{bid}.hnorm.weight",
+        "mtp.layers.0.eh_proj.weight":         "model.layers.{bid}.eh_proj.weight",
+        "mtp.layers.1.norm.weight":            "model.layers.{bid}.post_attention_layernorm.weight",
+        "mtp.layers.1.final_layernorm.weight": "model.layers.{bid}.shared_head.norm.weight",
+    }

+    def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
+        #   mtp.layers.0: NextN input fusion + attention
+        #   mtp.layers.1: MoE + final head norm
+        if self._mtp_bid is not None and name.startswith(("mtp.layers.0.", "mtp.layers.1.")):
+            suffix = name.split(".", 3)[3]
+            bid = self._mtp_bid
+            renamed = self._MTP_SPECIAL_RENAMES.get(name)
+            name = renamed.format(bid=bid) if renamed else f"backbone.layers.{bid}.{suffix}"
+
+        if self.is_moe and bid is not None:
            if name.endswith("mixer.gate.e_score_correction.bias"):
                yield from ModelBase.modify_tensors(self, data_torch, name, bid)
                return
@@ -2608,7 +2608,7 @@ static bool ggml_thread_apply_priority(int32_t prio) {
    return true;
 }

-#elif defined(__gnu_linux__)
+#elif defined(__linux__)
 // TODO: this may not work on BSD, to be verified

 static bool ggml_thread_apply_affinity(const bool * mask) {
@@ -5185,7 +5185,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
            return max_bias == 0.0f;
        }
        case GGML_OP_ROLL:
-            if(op->src[0]->type == GGML_TYPE_F32) {
+            if(op->src[0]->type == GGML_TYPE_F32 && ggml_is_contiguous(op->src[0])) {
                return true;
            }
            return false;
@@ -1268,8 +1268,9 @@ bool ggml_metal_device_supports_op(ggml_metal_device_t dev, const struct ggml_te
        case GGML_OP_ARGSORT:
        case GGML_OP_TOP_K:
        case GGML_OP_ARANGE:
-        case GGML_OP_ROLL:
            return true;
+        case GGML_OP_ROLL:
+            return ggml_is_contiguous(op->src[0]);
        case GGML_OP_FLASH_ATTN_EXT:
            // for new head sizes, add checks here
            if (op->src[0]->ne[0] != 32 &&
@@ -509,6 +509,7 @@ class MODEL_ARCH(IntEnum):
    OLMO             = auto()
    OLMO2            = auto()
    OLMOE            = auto()
+    MUSE_GLIMMER     = auto()
    OPENELM          = auto()
    ARCTIC           = auto()
    DEEPSEEK         = auto()
@@ -1181,6 +1182,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
    MODEL_ARCH.OLMO:             "olmo",
    MODEL_ARCH.OLMO2:            "olmo2",
    MODEL_ARCH.OLMOE:            "olmoe",
+    MODEL_ARCH.MUSE_GLIMMER:     "muse-glimmer",
    MODEL_ARCH.OPENELM:          "openelm",
    MODEL_ARCH.ARCTIC:           "arctic",
    MODEL_ARCH.DEEPSEEK:         "deepseek",
@@ -1562,8 +1564,8 @@ TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
    MODEL_TENSOR.V_MM_UP:                   "mm.up",
    MODEL_TENSOR.V_MM_DOWN:                 "mm.down",
    MODEL_TENSOR.V_MM_GATE:                 "mm.gate",
-    MODEL_TENSOR.V_MM_MERGER_FC1:            "mm.merger.fc1",
-    MODEL_TENSOR.V_MM_MERGER_FC2:            "mm.merger.fc2",
+    MODEL_TENSOR.V_MM_MERGER_FC1:           "mm.merger.fc1",
+    MODEL_TENSOR.V_MM_MERGER_FC2:           "mm.merger.fc2",
    MODEL_TENSOR.V_TOK_BOI:                 "v.boi",
    MODEL_TENSOR.V_TOK_EOI:                 "v.eoi",
    MODEL_TENSOR.V_MM_PRE_NORM:             "mm.pre_norm",
@@ -3331,6 +3333,25 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
        MODEL_TENSOR.FFN_UP_EXP,
        MODEL_TENSOR.FFN_DOWN_EXP,
    ],
+    MODEL_ARCH.MUSE_GLIMMER: [
+        MODEL_TENSOR.TOKEN_EMBD,
+        MODEL_TENSOR.OUTPUT,
+        MODEL_TENSOR.OUTPUT_NORM,
+        MODEL_TENSOR.ATTN_Q,
+        MODEL_TENSOR.ATTN_Q_NORM,
+        MODEL_TENSOR.ATTN_K,
+        MODEL_TENSOR.ATTN_K_NORM,
+        MODEL_TENSOR.ATTN_V,
+        MODEL_TENSOR.ATTN_OUT,
+        MODEL_TENSOR.ATTN_GATE,
+        MODEL_TENSOR.FFN_GATE,
+        MODEL_TENSOR.FFN_DOWN,
+        MODEL_TENSOR.FFN_UP,
+        MODEL_TENSOR.ATTN_NORM,
+        MODEL_TENSOR.ATTN_POST_NORM,
+        MODEL_TENSOR.FFN_PRE_NORM,
+        MODEL_TENSOR.FFN_POST_NORM,
+    ],
    MODEL_ARCH.OPENELM: [
        MODEL_TENSOR.TOKEN_EMBD,
        MODEL_TENSOR.OUTPUT_NORM,
@@ -3846,6 +3867,12 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
        MODEL_TENSOR.FFN_DOWN_SHEXP,
        MODEL_TENSOR.FFN_UP_SHEXP,
        MODEL_TENSOR.FFN_EXP_PROBS_B,
+        # NextN/MTP (draft head)
+        MODEL_TENSOR.ATTN_POST_NORM,
+        MODEL_TENSOR.NEXTN_EH_PROJ,
+        MODEL_TENSOR.NEXTN_ENORM,
+        MODEL_TENSOR.NEXTN_HNORM,
+        MODEL_TENSOR.NEXTN_SHARED_HEAD_NORM,
    ],
    MODEL_ARCH.EXAONE: [
        MODEL_TENSOR.TOKEN_EMBD,
@@ -5160,6 +5187,7 @@ class VisionProjectorType:
    MIMOVL         = "mimovl"
    MIMO_AUDIO     = "mimo_audio"
    GRANITE4_VISION = "granite4_vision"
+    MUSE_GLIMMER   = "muse-glimmer"


 # Items here are (block size, type size)
@@ -382,7 +382,7 @@ class TensorNameMap:
        ),

        MODEL_TENSOR.ATTN_GATE: (
-            "model.layers.{bid}.self_attn.gate_proj", # afmoe
+            "model.layers.{bid}.self_attn.gate_proj", # afmoe muse-glimmer
            "model.layers.{bid}.linear_attn.in_proj_z",  # qwen3.5
            "model.layers.{bid}.self_attn.g_proj",    # step3.5 head-wise attention gate
        ),
@@ -1298,10 +1298,12 @@ class TensorNameMap:
            "encoder.final_layer_norm", # t5
            "layer_norm",               # neobert
            "model.hidden_norm",        # dflash
+            "encoder.output_norm_enc",  # dflash (transformers MuseGlimmerAssistant)
        ),

        MODEL_TENSOR.FC: (
-            "model.fc", # dflash
+            "model.fc",   # dflash
+            "encoder.fc", # dflash (transformers MuseGlimmerAssistant)
        ),

        MODEL_TENSOR.DSPARK_MARKOV_W1: (
@@ -1467,6 +1469,7 @@ class TensorNameMap:
            "vision_tower.patch_embed.patchifier.proj", # dots.ocr
            "vision_model.conv1", # Step3-VL
            "model.vision_embedder.patch_dense", # gemma4 unified
+            "model.vision_tower.patch_embedder.patch_embedding", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_EMBD_NORM: (
@@ -1534,7 +1537,8 @@ class TensorNameMap:
            "siglip2.vision_model.encoder.layers.{bid}.self_attn.q_proj", # youtuvl
            "model.vision_model.transformer.layers.{bid}.self_attn.q_proj", # Deepseek-OCR CLIP, generated
            "vision_model.model.layers.{bid}.self_attn.q_proj.linear", # gemma4
-            "model.qwen2_model.model.model.layers.{bid}.self_attn.q_proj" # Deepseek-OCR-2 qwen2
+            "model.qwen2_model.model.model.layers.{bid}.self_attn.q_proj", # Deepseek-OCR-2 qwen2
+            "model.vision_tower.layers.{bid}.attn.q_proj", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_ATTN_Q_NORM: (
@@ -1560,7 +1564,8 @@ class TensorNameMap:
            "model.vision_model.transformer.layers.{bid}.self_attn.k_proj", # Deepseek-OCR CLIP, generated
            "siglip2.vision_model.encoder.layers.{bid}.self_attn.k_proj",
            "vision_model.model.layers.{bid}.self_attn.k_proj.linear", # gemma4
-            "model.qwen2_model.model.model.layers.{bid}.self_attn.k_proj" # Deepseek-OCR-2 qwen2
+            "model.qwen2_model.model.model.layers.{bid}.self_attn.k_proj", # Deepseek-OCR-2 qwen2
+            "model.vision_tower.layers.{bid}.attn.k_proj", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_ATTN_K_NORM: (
@@ -1586,7 +1591,8 @@ class TensorNameMap:
            "siglip2.vision_model.encoder.layers.{bid}.self_attn.v_proj",
            "model.vision_model.transformer.layers.{bid}.self_attn.v_proj", # Deepseek-OCR CLIP, generated
            "vision_model.model.layers.{bid}.self_attn.v_proj.linear", # gemma4
-            "model.qwen2_model.model.model.layers.{bid}.self_attn.v_proj" # Deepseek-OCR-2 qwen2
+            "model.qwen2_model.model.model.layers.{bid}.self_attn.v_proj", # Deepseek-OCR-2 qwen2
+            "model.vision_tower.layers.{bid}.attn.v_proj", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_INPUT_NORM: (
@@ -1610,6 +1616,7 @@ class TensorNameMap:
            "vision_tower.blocks.{bid}.norm1", # dots.ocr
            "vision_model.transformer.resblocks.{bid}.ln_1", # Step3-VL
            "model.qwen2_model.model.model.layers.{bid}.input_layernorm", # Deepseek-OCR-2 qwen2
+            "model.vision_tower.layers.{bid}.norm1", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_ATTN_O: (
@@ -1635,6 +1642,7 @@ class TensorNameMap:
            "vision_model.model.layers.{bid}.self_attn.o_proj.linear", # gemma4
            "vision_tower.blocks.{bid}.attn.proj", # dots.ocr
            "vision_model.transformer.resblocks.{bid}.attn.out_proj", # Step3-VL
+            "model.vision_tower.layers.{bid}.attn.proj", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_ATTN_SINKS: (
@@ -1663,6 +1671,7 @@ class TensorNameMap:
            "vision_tower.blocks.{bid}.norm2", # dots.ocr
            "vision_model.transformer.resblocks.{bid}.ln_2", # Step3-VL
            "model.qwen2_model.model.model.layers.{bid}.post_attention_layernorm", # Deepseek-OCR-2 qwen2
+            "model.vision_tower.layers.{bid}.norm2", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_FFN_UP: (
@@ -1687,6 +1696,7 @@ class TensorNameMap:
            "vision_model.model.layers.{bid}.mlp.up_proj", # gemma4
            "vision_model.transformer.resblocks.{bid}.mlp.c_fc", # Step3-VL
            "model.qwen2_model.model.model.layers.{bid}.mlp.up_proj", # Deepseek-OCR-2 qwen2
+            "model.vision_tower.layers.{bid}.mlp.fc1", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_FFN_GATE: (
@@ -1719,6 +1729,7 @@ class TensorNameMap:
            "model.qwen2_model.model.model.layers.{bid}.mlp.down_proj" , # Deepseek-OCR-2 qwen2
            "vision_model.model.layers.{bid}.mlp.down_proj", # gemma4
            "vision_model.transformer.resblocks.{bid}.mlp.c_proj", # Step3-VL
+            "model.vision_tower.layers.{bid}.mlp.fc2", # muse-glimmer
        ),

        MODEL_TENSOR.V_ENC_ATTN_POST_NORM: (
@@ -1753,6 +1764,7 @@ class TensorNameMap:
            "model.vision_model.pre_layrnorm", # Deepseek-OCR CLIP
            "vision_tower.patch_embed.patchifier.norm", # dots.ocr
            "vision_model.ln_pre", # Step3-VL
+            "model.vision_tower.ln_pre", # muse-glimmer
        ),

        MODEL_TENSOR.V_POST_NORM: (
@@ -1766,6 +1778,7 @@ class TensorNameMap:
            "visual.post_layernorm", # glm4v
            "siglip2.vision_model.post_layernorm",
            "model.qwen2_model.model.model.norm", # Deepseek-OCR-2 qwen2
+            "model.vision_tower.ln_post", # muse-glimmer
        ),

        MODEL_TENSOR.V_MM_POST_NORM: (
@@ -1,8 +1,9 @@
 {#- Iteration on laguna_glm_thinking_v8/chat_template.jinja -#}
 {#- No formatting instructions -#}
 {{- "〈|EOS|〉" -}}
-{%- set enable_thinking = enable_thinking | default(false) -%}
+{%- set enable_thinking = enable_thinking | default(true) -%}
 {%- set add_generation_prompt = add_generation_prompt | default(false) -%}
+{%- set preserve_thinking = preserve_thinking | default(false) -%}

 {#- ───── header (system message) ───── -#}
 {#- A caller-supplied system message with empty content opts out of the default below, producing no <system> block — used to train without a system message. -#}
@@ -51,7 +52,7 @@
        {%- set reasoning_content = message.reasoning_content -%}
      {%- endif -%}
      {#- Display reasoning content for all messages if enable_thinking -#}
-      {%- if enable_thinking -%}
+      {%- if enable_thinking or preserve_thinking -%}
        {{- '<think>' + reasoning_content + '</think>' -}}
      {%- else -%}
        {{- '</think>' -}}
@@ -5,7 +5,7 @@ import os
 import sys
 import subprocess

-HTTPLIB_VERSION = "refs/tags/v0.52.0"
+HTTPLIB_VERSION = "refs/tags/v0.53.0"

 vendor = {
    "https://github.com/nlohmann/json/releases/latest/download/json.hpp":     "vendor/nlohmann/json.hpp",
@@ -21,34 +21,13 @@ vendor = {
    f"https://raw.githubusercontent.com/yhirose/cpp-httplib/{HTTPLIB_VERSION}/split.py":  "split.py",
    f"https://raw.githubusercontent.com/yhirose/cpp-httplib/{HTTPLIB_VERSION}/LICENSE":   "vendor/cpp-httplib/LICENSE",

-    "https://raw.githubusercontent.com/sheredom/subprocess.h/8671cee1fc09f11a70ce3782a0ee13177c3aa387/subprocess.h": "vendor/sheredom/subprocess.h",
+    "https://raw.githubusercontent.com/sheredom/subprocess.h/9ce0d701b6fb10f8f8c4445edd31e7c60a1237e3/subprocess.h": "vendor/sheredom/subprocess.h",
 }

-# TODO @ngxson : this is temporary, to be removed in the future
-patches = [
-    # https://github.com/sheredom/subprocess.h/pull/102
-    "vendor/sheredom/patch-bsd.patch",
-    # https://github.com/sheredom/subprocess.h/pull/101
-    "vendor/sheredom/patch-windows-quote-backslash.patch",
-    # https://github.com/sheredom/subprocess.h/pull/104
-    # note: must be applied after patch-bsd.patch, they touch adjacent lines
-    "vendor/sheredom/patch-glibc-older-than-2.29.patch",
-]
-
 for url, filename in vendor.items():
    print(f"downloading {url} to {filename}") # noqa: NP100
    urllib.request.urlretrieve(url, filename)

-for patch in patches:
-    print(f"applying {patch}") # noqa: NP100
-    try:
-        subprocess.check_call([
-            "git", "apply", "--directory", os.path.dirname(patch), patch
-        ])
-    except Exception as e:
-        print(f"Error: {e}") # noqa: NP100
-        sys.exit(1)
-
 print("Splitting httplib.h...") # noqa: NP100
 try:
    subprocess.check_call([
@@ -71,6 +71,7 @@ static const std::map<llm_arch, const char *> LLM_ARCH_NAMES = {
    { LLM_ARCH_OLMO,             "olmo"             },
    { LLM_ARCH_OLMO2,            "olmo2"            },
    { LLM_ARCH_OLMOE,            "olmoe"            },
+    { LLM_ARCH_MUSE_GLIMMER,     "muse-glimmer"     },
    { LLM_ARCH_OPENELM,          "openelm"          },
    { LLM_ARCH_ARCTIC,           "arctic"           },
    { LLM_ARCH_DEEPSEEK,         "deepseek"         },
@@ -76,6 +76,7 @@ enum llm_arch {
    LLM_ARCH_OLMO,
    LLM_ARCH_OLMO2,
    LLM_ARCH_OLMOE,
+    LLM_ARCH_MUSE_GLIMMER,
    LLM_ARCH_OPENELM,
    LLM_ARCH_ARCTIC,
    LLM_ARCH_DEEPSEEK,
@@ -27,6 +27,7 @@ bool llama_model_saver_supports_arch(llm_arch arch) {
        case LLM_ARCH_APERTUS:
        case LLM_ARCH_MIMO2:
        case LLM_ARCH_STEP35:
+        case LLM_ARCH_MUSE_GLIMMER:
        case LLM_ARCH_MELLUM:
        case LLM_ARCH_LAGUNA:
            return false;
@@ -40,6 +40,8 @@

 static llama_model * llama_model_mapping(llm_arch arch, const llama_model_params & params) {
    switch (arch) {
+        case LLM_ARCH_CLIP:
+            return new llama_model_clip(params);
        case LLM_ARCH_LLAMA:
            return new llama_model_llama(params);
        case LLM_ARCH_LLAMA4:
@@ -174,6 +176,8 @@ static llama_model * llama_model_mapping(llm_arch arch, const llama_model_params
            return new llama_model_olmo2(params);
        case LLM_ARCH_OLMOE:
            return new llama_model_olmoe(params);
+        case LLM_ARCH_MUSE_GLIMMER:
+            return new llama_model_muse_glimmer(params);
        case LLM_ARCH_OPENELM:
            return new llama_model_openelm(params);
        case LLM_ARCH_GPTNEOX:
@@ -2231,6 +2235,9 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
                    params.ctx_type == LLAMA_CONTEXT_TYPE_MTP &&
                    (arch == LLM_ARCH_QWEN3NEXT || arch == LLM_ARCH_QWEN35 || arch == LLM_ARCH_QWEN35MOE);

+                const bool mtp_on_hybrid_nemotron =
+                    params.ctx_type == LLAMA_CONTEXT_TYPE_MTP && arch == LLM_ARCH_NEMOTRON_H_MOE;
+
                if (llm_arch_is_recurrent(arch)) {
                    res = new llama_memory_recurrent(
                            *this,
@@ -2241,7 +2248,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
                            cparams.n_seq_max,
                            cparams.n_rs_seq,
                            nullptr);
-                } else if (llm_arch_is_hybrid(arch) && !mtp_on_hybrid_qwen) {
+                } else if (llm_arch_is_hybrid(arch) && !mtp_on_hybrid_qwen && !mtp_on_hybrid_nemotron) {
                    // The main difference between hybrid architectures is the
                    // layer filters, so pick the right one here
                    llama_memory_hybrid::layer_filter_cb filter_attn = nullptr;
@@ -2322,7 +2329,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
                        };
                    }

-                    if (mtp_on_hybrid_qwen) {
+                    if (mtp_on_hybrid_qwen || mtp_on_hybrid_nemotron) {
                        filter = [&](uint32_t il) { return il >= hparams.n_layer(); };
                    }

@@ -2594,6 +2601,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
        case LLM_ARCH_DEEPSEEK2OCR:
        case LLM_ARCH_DEEPSEEK32:
        case LLM_ARCH_DEEPSEEK4:
+        case LLM_ARCH_MUSE_GLIMMER:
        case LLM_ARCH_PLM:
        case LLM_ARCH_CHATGLM:
        case LLM_ARCH_GRANITE:
@@ -0,0 +1,18 @@
+#include "models.h"
+
+// Stub to allow llama-quantize to open mmproj GGUFs
+
+[[noreturn]]
+void llama_model_clip::load_arch_hparams(llama_model_loader &) {
+    GGML_ABORT("CLIP is a quant-only stub; load_arch_hparams should not be called");
+}
+
+[[noreturn]]
+void llama_model_clip::load_arch_tensors(llama_model_loader &) {
+    GGML_ABORT("CLIP is a quant-only stub; load_arch_tensors should not be called");
+}
+
+[[noreturn]]
+std::unique_ptr<llm_graph_context> llama_model_clip::build_arch_graph(const llm_graph_params &) const {
+    GGML_ABORT("CLIP has no inference graph via llama_model dispatch; runtime lives in tools/mtmd/clip.cpp");
+}
@@ -386,6 +386,22 @@ struct llama_model_bloom : public llama_model_base {
 };


+// Quant-only stub for mmproj GGUFs
+// none of these are ever called, they only exist to satisfy the llama_model_base interface
+struct llama_model_clip : public llama_model_base {
+    llama_model_clip(const struct llama_model_params & params) : llama_model_base(params) {}
+
+    [[noreturn]]
+    void load_arch_hparams(llama_model_loader & ml) override;
+
+    [[noreturn]]
+    void load_arch_tensors(llama_model_loader & ml) override;
+
+    [[noreturn]]
+    std::unique_ptr<llm_graph_context> build_arch_graph(const llm_graph_params & params) const override;
+};
+
+
 struct llama_model_mpt : public llama_model_base {
    llama_model_mpt(const struct llama_model_params & params) : llama_model_base(params) {}
    void load_arch_hparams(llama_model_loader & ml) override;
@@ -1028,6 +1044,19 @@ struct llama_model_olmoe : public llama_model_base {
 };


+struct llama_model_muse_glimmer : public llama_model_base {
+    llama_model_muse_glimmer(const struct llama_model_params & params) : llama_model_base(params) {}
+    void load_arch_hparams(llama_model_loader & ml) override;
+    void load_arch_tensors(llama_model_loader & ml) override;
+
+    struct graph : public llm_graph_context {
+        graph(const llama_model & model, const llm_graph_params & params);
+    };
+
+    std::unique_ptr<llm_graph_context> build_arch_graph(const llm_graph_params & params) const override;
+};
+
+
 struct llama_model_openelm : public llama_model_base {
    llama_model_openelm(const struct llama_model_params & params) : llama_model_base(params) {}
    void load_arch_hparams(llama_model_loader & ml) override;
@@ -1461,6 +1490,10 @@ struct llama_model_nemotron_h_moe : public llama_model_nemotron_h {

    using graph = llama_model_nemotron_h::graph;

+    struct graph_mtp : public llm_graph_context {
+        graph_mtp(const llama_model & model, const llm_graph_params & params);
+    };
+
    std::unique_ptr<llm_graph_context> build_arch_graph(const llm_graph_params & params) const override;
 };

@@ -0,0 +1,208 @@
+#include "models.h"
+
+void llama_model_muse_glimmer::load_arch_hparams(llama_model_loader & ml) {
+    ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
+    ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW,    hparams.n_swa);
+    ml.get_key(LLM_KV_FINAL_LOGIT_SOFTCAPPING,     hparams.f_final_logit_softcapping, false);
+    ml.get_key(LLM_KV_LOGIT_SCALE,                 hparams.f_logit_scale);
+
+    hparams.rope_freq_base_train_swa = hparams.rope_freq_base_train;
+    ml.get_key(LLM_KV_ROPE_FREQ_BASE_SWA, hparams.rope_freq_base_train_swa, false);
+
+    hparams.swa_type = LLAMA_SWA_TYPE_STANDARD;
+    uint32_t swa_period = 4;
+    if (ml.get_key_or_arr(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, swa_period, false)) {
+        hparams.set_swa_pattern(swa_period);
+    } else {
+        ml.get_key_or_arr(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, hparams.is_swa_impl, hparams.n_layer());
+    }
+
+    switch (hparams.n_layer()) {
+        case 52: type = LLM_TYPE_30B; break;
+        default: type = LLM_TYPE_UNKNOWN;
+    }
+}
+
+void llama_model_muse_glimmer::load_arch_tensors(llama_model_loader &) {
+    LLAMA_LOAD_LOCALS;
+
+    tok_embd    = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD,  "weight"), {n_embd, n_vocab}, 0);
+    output_norm = create_tensor(tn(LLM_TENSOR_OUTPUT_NORM, "weight"), {n_embd}, 0);
+    output      = create_tensor(tn(LLM_TENSOR_OUTPUT,      "weight"), {n_embd, n_vocab}, 0);
+
+    for (int i = 0; i < n_layer; ++i) {
+        auto & layer = layers[i];
+
+        // Pre/post-attention norms (Muse Glimmer's `weight + 1` applied at conversion time).
+        layer.attn_norm      = create_tensor(tn(LLM_TENSOR_ATTN_NORM,      "weight", i), {n_embd}, 0);
+        layer.attn_post_norm = create_tensor(tn(LLM_TENSOR_ATTN_POST_NORM, "weight", i), {n_embd}, 0);
+
+        // Q/K/V/O projections.
+        create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head, n_embd_k_gqa, n_embd_v_gqa, 0);
+        layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_k * n_head, n_embd}, 0);
+
+        // QK-norm. Weights are synthesized at conversion time to absorb `qk_scale_factor`.
+        layer.attn_q_norm = create_tensor(tn(LLM_TENSOR_ATTN_Q_NORM, "weight", i), {n_embd_head_k}, 0);
+        layer.attn_k_norm = create_tensor(tn(LLM_TENSOR_ATTN_K_NORM, "weight", i), {n_embd_head_k}, 0);
+
+        // Attention output gate: sigmoid(gate) * attn_out before o_proj (same as afmoe).
+        layer.wqkv_gate = create_tensor(tn(LLM_TENSOR_ATTN_GATE, "weight", i), {n_embd, n_embd_head_k * n_head}, 0);
+
+        // Pre/post-FFN norms (FFN_PRE_NORM is aliased to LLM_TENSOR_FFN_NORM).
+        layer.ffn_norm      = create_tensor(tn(LLM_TENSOR_FFN_NORM,      "weight", i), {n_embd}, 0);
+        layer.ffn_post_norm = create_tensor(tn(LLM_TENSOR_FFN_POST_NORM, "weight", i), {n_embd}, 0);
+
+        // Dense FFN (unlike afmoe, no MoE branches).
+        layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd, n_ff}, 0);
+        layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {n_ff, n_embd}, 0);
+        layer.ffn_up   = create_tensor(tn(LLM_TENSOR_FFN_UP,   "weight", i), {n_embd, n_ff}, 0);
+    }
+}
+
+llama_model_muse_glimmer::graph::graph(const llama_model & model, const llm_graph_params & params)
+    : llm_graph_context(params) {
+    const int64_t n_embd_head = hparams.n_embd_head_v();
+    GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
+
+    // Different to f_norm_rms_eps for post-attn / post-FFN norms
+    const float post_norm_eps = 1e-8f;
+
+    ggml_tensor * cur;
+    ggml_tensor * inpL;
+
+    inpL = build_inp_embd(model.tok_embd);
+    inpL = build_norm(inpL, nullptr, nullptr, LLM_NORM_RMS, -1);
+    cb(inpL, "embd_norm", -1);
+
+    ggml_tensor * inp_pos = build_inp_pos();
+    auto * inp_attn = build_attn_inp_kv_iswa();
+    ggml_tensor * inp_out_ids = build_inp_out_ids();
+
+    const float kq_scale = 1.0f / sqrtf(float(n_embd_head));
+
+    for (int il = 0; il < n_layer; ++il) {
+        // expose per-layer residual for speculative drafts (see LLM_KV_TARGET_LAYERS).
+        res->t_layer_inp[il] = inpL;
+
+        const float freq_base_l  = model.get_rope_freq_base (cparams, il);
+        const float freq_scale_l = model.get_rope_freq_scale(cparams, il);
+
+        ggml_tensor * inpSA = inpL;
+
+        // RoPE runs on the SWA layers, NoPE on full ones.
+        const bool use_rope = hparams.is_swa(il);
+
+        // pre-attention norm (weight+1 folded at conversion time)
+        cur = build_norm(inpL, model.layers[il].attn_norm, NULL, LLM_NORM_RMS, il);
+        cb(cur, "attn_norm", il);
+
+        // self-attention: attention output gate around SDPA (afmoe.cpp:147-191)
+        {
+            ggml_tensor * attn_inp = cur;  // save input for gate computation
+
+            auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
+                    n_embd_head, n_head, n_head_kv, il);
+
+            // gate = wqkv_gate @ attn_inp (from pre-attn hidden state)
+            ggml_tensor * gate = build_lora_mm(model.layers[il].wqkv_gate, attn_inp);
+            cb(gate, "attn_gate_proj", il);
+
+            // QK-norm. attn_q_norm weight was synthesized at conversion to broadcast
+            // qk_scale_factor across head_dim; attn_k_norm is identity (ones).
+            Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
+            Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
+            cb(Qcur, "Qcur_normed", il);
+            cb(Kcur, "Kcur_normed", il);
+
+            if (use_rope) {
+                Qcur = ggml_rope_ext(
+                        ctx0, Qcur, inp_pos, nullptr,
+                        n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
+                        ext_factor, attn_factor, beta_fast, beta_slow);
+                cb(Qcur, "Qcur_rope", il);
+
+                Kcur = ggml_rope_ext(
+                        ctx0, Kcur, inp_pos, nullptr,
+                        n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
+                        ext_factor, attn_factor, beta_fast, beta_slow);
+                cb(Kcur, "Kcur_rope", il);
+            }
+
+            // SDPA. wo is deferred; the gate goes between attn_out and o_proj.
+            cur = build_attn(inp_attn,
+                    NULL, NULL, NULL,
+                    Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
+            cb(cur, "attn_out", il);
+
+            gate = ggml_sigmoid(ctx0, gate);
+            cb(gate, "attn_gate_sig", il);
+            cur = ggml_mul(ctx0, cur, gate);
+            cb(cur, "attn_gated", il);
+
+            cur = build_lora_mm(model.layers[il].wo, cur, model.layers[il].wo_s);
+            cb(cur, "attn_o_proj", il);
+        }
+
+        cur = ggml_rms_norm(ctx0, cur, post_norm_eps);
+        cur = ggml_mul(ctx0, cur, model.layers[il].attn_post_norm);
+        cb(cur, "attn_post_norm", il);
+
+        if (il == n_layer - 1 && inp_out_ids) {
+            cur   = ggml_get_rows(ctx0,   cur, inp_out_ids);
+            inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
+        }
+
+        ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpSA);
+        cb(ffn_inp, "ffn_inp", il);
+
+        // pre-FFN norm
+        cur = build_norm(ffn_inp, model.layers[il].ffn_norm, NULL, LLM_NORM_RMS, il);
+        cb(cur, "ffn_norm", il);
+
+        // SwiGLU dense FFN
+        cur = build_ffn(cur,
+                model.layers[il].ffn_up,   NULL, NULL,
+                model.layers[il].ffn_gate, NULL, NULL,
+                model.layers[il].ffn_down, NULL, NULL,
+                NULL,
+                LLM_FFN_SILU, LLM_FFN_PAR, il);
+        cb(cur, "ffn_out", il);
+
+        cur = ggml_rms_norm(ctx0, cur, post_norm_eps);
+        cur = ggml_mul(ctx0, cur, model.layers[il].ffn_post_norm);
+        cb(cur, "ffn_post_norm", il);
+
+        cur = ggml_add(ctx0, cur, ffn_inp);
+        cur = build_cvec(cur, il);
+        cb(cur, "l_out", il);
+
+        inpL = cur;
+    }
+
+    cur = inpL;
+
+    // final norm
+    cur = build_norm(cur, model.output_norm, NULL, LLM_NORM_RMS, -1);
+    cb(cur, "result_norm", -1);
+    res->t_embd = cur;
+
+    // lm_head, followed by output multiplier
+    cur = build_lora_mm(model.output, cur, model.output_s);
+    cur = ggml_scale(ctx0, cur, hparams.f_logit_scale);
+
+    // Final logit tanh softcap (from gemma3.cpp).
+    if (hparams.f_final_logit_softcapping) {
+        cur = ggml_scale(ctx0, cur, 1.0f / hparams.f_final_logit_softcapping);
+        cur = ggml_tanh(ctx0, cur);
+        cur = ggml_scale(ctx0, cur, hparams.f_final_logit_softcapping);
+    }
+
+    cb(cur, "result_output", -1);
+    res->t_logits = cur;
+
+    ggml_build_forward_expand(gf, cur);
+}
+
+std::unique_ptr<llm_graph_context> llama_model_muse_glimmer::build_arch_graph(const llm_graph_params & params) const {
+    return std::make_unique<graph>(*this, params);
+}
@@ -1,6 +1,156 @@
 #include "models.h"

 std::unique_ptr<llm_graph_context> llama_model_nemotron_h_moe::build_arch_graph(const llm_graph_params & params) const {
+    if (params.gtype == LLM_GRAPH_TYPE_DECODER_MTP) {
+        return std::make_unique<graph_mtp>(*this, params);
+    }
    return std::make_unique<graph>(*this, params);
 }

+// MTP draft head for Nemotron-H MoE
+llama_model_nemotron_h_moe::graph_mtp::graph_mtp(const llama_model & model, const llm_graph_params & params)
+    : llm_graph_context(params) {
+    GGML_ASSERT(hparams.n_layer_nextn == 1 && "NEMOTRON_H_MOE MTP currently supports a single MTP block");
+
+    const int64_t n_embd_head = hparams.n_embd_head_v();
+    GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
+
+    const int il = hparams.n_layer();
+    const auto & layer = model.layers[il];
+
+    GGML_ASSERT(layer.nextn.eh_proj && layer.nextn.enorm && layer.nextn.hnorm);
+    GGML_ASSERT(layer.ffn_gate_inp);
+
+    // token embedding weights
+    ggml_tensor * tok_embd_w = layer.nextn.embed_tokens ? layer.nextn.embed_tokens : model.tok_embd;
+    GGML_ASSERT(tok_embd_w != nullptr && "NEMOTRON_H_MOE MTP requires token embeddings");
+
+    auto inp = std::make_unique<llm_graph_input_embd_h>(hparams.n_embd);
+
+    inp->tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_tokens);
+    ggml_set_input(inp->tokens);
+
+    inp->embd = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, hparams.n_embd_inp(), n_tokens);
+    ggml_set_input(inp->embd);
+
+    ggml_tensor * tok_embd;
+    if (ubatch.token) {
+        tok_embd = ggml_get_rows(ctx0, tok_embd_w, inp->tokens);
+    } else {
+        tok_embd = inp->embd;
+    }
+    cb(tok_embd, "mtp_tok_embd", il);
+
+    inp->h = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, hparams.n_embd, n_tokens);
+    ggml_set_input(inp->h);
+    ggml_set_name(inp->h, "mtp_h_input");
+
+    ggml_tensor * h_embd = inp->h;
+
+    res->add_input(std::move(inp));
+
+    ggml_tensor * inp_out_ids = build_inp_out_ids();
+
+    // attention fills KV over all tokens, but the MoE is position-wise: gather output rows before
+    // it to save FFN compute (unless unmasked embeddings_nextn needs the full-length hidden state)
+    const bool emit_h_nextn    = cparams.embeddings_nextn;
+    const bool crop_before_ffn = inp_out_ids && (!emit_h_nextn || cparams.embeddings_nextn_masked);
+
+    auto * inp_attn = build_attn_inp_kv();
+
+    ggml_tensor * h_norm = build_norm(h_embd, layer.nextn.hnorm, nullptr, LLM_NORM_RMS, il);
+    cb(h_norm, "mtp_hnorm", il);
+    ggml_tensor * e_norm = build_norm(tok_embd, layer.nextn.enorm, nullptr, LLM_NORM_RMS, il);
+    cb(e_norm, "mtp_enorm", il);
+
+    ggml_tensor * concat = ggml_concat(ctx0, e_norm, h_norm, /*dim=*/ 0);
+    cb(concat, "mtp_concat", il);
+
+    ggml_tensor * cur = build_lora_mm(layer.nextn.eh_proj, concat, layer.nextn.eh_proj_s);
+    cb(cur, "mtp_eh_proj", il);
+
+    // dense NoPE attention sub-layer (mtp.layers.0)
+    ggml_tensor * inpSA = cur;
+    cur = build_norm(cur, layer.attn_norm, nullptr, LLM_NORM_RMS, il);
+    cb(cur, "mtp_attn_norm", il);
+
+    {
+        auto [Qcur, Kcur, Vcur] = build_qkv(layer, cur, n_embd_head, hparams.n_head(il), hparams.n_head_kv(il), il);
+        const float kq_scale = hparams.f_attention_scale == 0.0f
+            ? 1.0f / sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
+        cur = build_attn(inp_attn, layer.wo, layer.wo_b, layer.wo_s,
+                Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
+        cb(cur, "mtp_attn_out", il);
+    }
+
+    cur = ggml_add(ctx0, cur, inpSA);
+    cb(cur, "mtp_attn_residual", il);
+
+    // gather the output rows here so the MoE FFN below only runs on the positions we keep
+    if (crop_before_ffn) {
+        cur = ggml_get_rows(ctx0, cur, inp_out_ids);
+    }
+
+    // MoE FFN sub-layer (mtp.layers.1)
+    ggml_tensor * ffn_residual = cur;
+    cur = build_norm(cur, layer.attn_post_norm, nullptr, LLM_NORM_RMS, il);
+    cb(cur, "mtp_attn_post_norm", il);
+
+    {
+        ggml_tensor * router_logits = build_lora_mm(layer.ffn_gate_inp, cur);
+        cb(router_logits, "mtp_ffn_moe_logits", il);
+
+        ggml_tensor * moe_out =
+            build_moe_ffn(cur,
+                layer.ffn_gate_inp,
+                layer.ffn_up_exps,
+                nullptr, // no gate
+                layer.ffn_down_exps,
+                layer.ffn_exp_probs_b,
+                n_expert, n_expert_used,
+                LLM_FFN_RELU_SQR, hparams.expert_weights_norm,
+                hparams.expert_weights_scale,
+                LLAMA_EXPERT_GATING_FUNC_TYPE_SIGMOID,
+                il,
+                router_logits, nullptr,
+                layer.ffn_up_exps_s,
+                nullptr, // no gate
+                layer.ffn_down_exps_s);
+        cb(moe_out, "mtp_ffn_moe_out", il);
+
+        ggml_tensor * ffn_shexp = build_ffn(cur,
+                layer.ffn_up_shexp,   NULL, layer.ffn_up_shexp_s,
+                NULL,                 NULL, NULL,
+                layer.ffn_down_shexp, NULL, layer.ffn_down_shexp_s,
+                NULL,
+                LLM_FFN_RELU_SQR, LLM_FFN_PAR, il);
+        cb(ffn_shexp, "mtp_ffn_shexp", il);
+
+        cur = ggml_add(ctx0, moe_out, ffn_shexp);
+        cb(cur, "mtp_ffn_out", il);
+    }
+
+    cur = ggml_add(ctx0, cur, ffn_residual);
+    cb(cur, "mtp_post_ffn", il);
+
+    // final head norm: the MTP head has its own LayerNorm
+    GGML_ASSERT(layer.nextn.shared_head_norm && "NEMOTRON_H_MOE MTP: missing final head norm");
+    cur = build_norm(cur, layer.nextn.shared_head_norm, nullptr, LLM_NORM, -1);
+
+    cb(cur, "h_nextn", -1);
+    res->t_h_nextn = cur;
+
+    if (!crop_before_ffn && inp_out_ids) {
+        cur = ggml_get_rows(ctx0, cur, inp_out_ids);
+    }
+
+    // LM head
+    ggml_tensor * head_w = layer.nextn.shared_head_head ? layer.nextn.shared_head_head : model.output;
+    ggml_tensor * head_s = layer.nextn.shared_head_head ? layer.nextn.shared_head_head_s : model.output_s;
+    GGML_ASSERT(head_w != nullptr && "NEMOTRON_H_MOE MTP requires an output projection");
+    cur = build_lora_mm(head_w, cur, head_s);
+    cb(cur, "result_output", -1);
+
+    res->t_logits = cur;
+    ggml_build_forward_expand(gf, cur);
+}
@@ -7,13 +7,18 @@ void llama_model_nemotron_h::load_arch_hparams(llama_model_loader & ml) {
    ml.get_key(LLM_KV_SSM_TIME_STEP_RANK, hparams.ssm_dt_rank);
    ml.get_key(LLM_KV_SSM_GROUP_COUNT,    hparams.ssm_n_group);

+    // NextN/MTP: optional draft head appended as extra trailing block(s)
+    ml.get_key(LLM_KV_NEXTN_PREDICT_LAYERS, hparams.n_layer_nextn, false);
+    GGML_ASSERT(hparams.n_layer_nextn < hparams.n_layer_all && "n_layer_nextn must be < n_layer_all");
+
    // A layer is recurrent IFF the n_head_kv value is set to 0 and
-    // the n_ff value is set to 0
-    for (uint32_t i = 0; i < hparams.n_layer(); ++i) {
-        hparams.is_recr_impl[i] = (hparams.n_head_kv(i) == 0 && hparams.n_ff(i) == 0);
+    // the n_ff value is set to 0. Appended MTP blocks are dense (non-recurrent)
+    for (uint32_t i = 0; i < hparams.n_layer_all; ++i) {
+        hparams.is_recr_impl[i] = i < hparams.n_layer() && hparams.n_head_kv(i) == 0 && hparams.n_ff(i) == 0;
    }

    ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
+    ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS,     hparams.f_norm_eps); // MTP head final_layernorm

    ml.get_key(LLM_KV_EXPERT_FEED_FORWARD_LENGTH,        hparams.n_ff_exp,        false);
    ml.get_key(LLM_KV_EXPERT_SHARED_FEED_FORWARD_LENGTH, hparams.n_ff_shexp,      false);
@@ -30,9 +35,13 @@ void llama_model_nemotron_h::load_arch_hparams(llama_model_loader & ml) {
    }
 }

-void llama_model_nemotron_h::load_arch_tensors(llama_model_loader &) {
+void llama_model_nemotron_h::load_arch_tensors(llama_model_loader & ml) {
    LLAMA_LOAD_LOCALS;

+    const bool mtp_only    = hparams.n_layer_nextn > 0 && ml.get_weight("blk.0.attn_norm.weight") == nullptr;
+    const int  trunk_flags = mtp_only ? TENSOR_NOT_REQUIRED : 0;
+    const int  mtp_flags   = !ml.load_mtp ? TENSOR_SKIP : 0;
+
    // mamba2 Mixer SSM params
    // NOTE: int64_t for tensor dimensions
    const int64_t d_conv     = hparams.ssm_d_conv;
@@ -60,61 +69,94 @@ void llama_model_nemotron_h::load_arch_tensors(llama_model_loader &) {
        auto & layer = layers[i];

        // all blocks use the attn norm
-        layer.attn_norm  = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, 0);
+        layer.attn_norm  = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, trunk_flags);

        if (hparams.is_recr(i)) {
            // ssm layers
-            layer.ssm_in = create_tensor(tn(LLM_TENSOR_SSM_IN, "weight", i), {n_embd, d_in_proj}, 0);
+            layer.ssm_in = create_tensor(tn(LLM_TENSOR_SSM_IN, "weight", i), {n_embd, d_in_proj}, trunk_flags);

-            layer.ssm_conv1d = create_tensor(tn(LLM_TENSOR_SSM_CONV1D, "weight", i), {d_conv, d_inner + 2*n_group*d_state}, 0);
+            layer.ssm_conv1d = create_tensor(tn(LLM_TENSOR_SSM_CONV1D, "weight", i), {d_conv, d_inner + 2*n_group*d_state}, trunk_flags);
            layer.ssm_conv1d_b = create_tensor(tn(LLM_TENSOR_SSM_CONV1D, "bias", i), {d_inner + 2*n_group*d_state}, TENSOR_NOT_REQUIRED);

-            layer.ssm_dt_b = create_tensor(tn(LLM_TENSOR_SSM_DT, "bias", i), {n_ssm_head}, 0);
+            layer.ssm_dt_b = create_tensor(tn(LLM_TENSOR_SSM_DT, "bias", i), {n_ssm_head}, trunk_flags);

            // no "weight" suffix for these
-            layer.ssm_a = create_tensor(tn(LLM_TENSOR_SSM_A, i), {1, n_ssm_head}, 0);
-            layer.ssm_d = create_tensor(tn(LLM_TENSOR_SSM_D, i), {1, n_ssm_head}, 0);
+            layer.ssm_a = create_tensor(tn(LLM_TENSOR_SSM_A, i), {1, n_ssm_head}, trunk_flags);
+            layer.ssm_d = create_tensor(tn(LLM_TENSOR_SSM_D, i), {1, n_ssm_head}, trunk_flags);

-            layer.ssm_norm = create_tensor(tn(LLM_TENSOR_SSM_NORM, "weight", i), {d_inner / n_group, n_group}, 0);
+            layer.ssm_norm = create_tensor(tn(LLM_TENSOR_SSM_NORM, "weight", i), {d_inner / n_group, n_group}, trunk_flags);

            // out_proj
-            layer.ssm_out = create_tensor(tn(LLM_TENSOR_SSM_OUT, "weight", i), {d_inner, n_embd}, 0);
+            layer.ssm_out = create_tensor(tn(LLM_TENSOR_SSM_OUT, "weight", i), {d_inner, n_embd}, trunk_flags);
        } else if (hparams.n_ff(i) == 0) {
            // attention layers (with optional bias)
            const int64_t n_head_i = hparams.n_head(i);
            const int64_t n_embd_k_gqa_i = hparams.n_embd_k_gqa(i);
            const int64_t n_embd_v_gqa_i = hparams.n_embd_v_gqa(i);
-            create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head_i, n_embd_k_gqa_i, n_embd_v_gqa_i, 0);
-            layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_k * n_head_i, n_embd}, 0);
+            create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head_i, n_embd_k_gqa_i, n_embd_v_gqa_i, trunk_flags);
+            layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_k * n_head_i, n_embd}, trunk_flags);
            layer.wo_b = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "bias", i), {n_embd}, TENSOR_NOT_REQUIRED);
        }  else {
            if (n_expert != 0) {
                const int64_t n_ff_exp = hparams.n_ff_exp ? hparams.n_ff_exp : n_ff / n_expert_used;
                const int64_t n_ff_shexp = hparams.n_ff_shexp;

-                layer.ffn_gate_inp    = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP,  "weight", i), { n_embd, n_expert}, 0);
-                layer.ffn_exp_probs_b = create_tensor(tn(LLM_TENSOR_FFN_EXP_PROBS_B, "bias", i), {n_expert         }, 0);
+                layer.ffn_gate_inp    = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP,  "weight", i), { n_embd, n_expert}, trunk_flags);
+                layer.ffn_exp_probs_b = create_tensor(tn(LLM_TENSOR_FFN_EXP_PROBS_B, "bias", i), {n_expert         }, trunk_flags);

                // MoE branch
                layer.ffn_latent_down = create_tensor(tn(LLM_TENSOR_FFN_LATENT_DOWN, "weight", i), {n_embd, moe_n_embd}, TENSOR_NOT_REQUIRED);
                layer.ffn_latent_up   = create_tensor(tn(LLM_TENSOR_FFN_LATENT_UP,   "weight", i), {moe_n_embd, n_embd}, TENSOR_NOT_REQUIRED);

-                layer.ffn_down_exps   = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), {n_ff_exp,   moe_n_embd, n_expert}, 0);
-                layer.ffn_up_exps     = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS,   "weight", i), {moe_n_embd, n_ff_exp, n_expert}, 0);
+                layer.ffn_down_exps   = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), {n_ff_exp,   moe_n_embd, n_expert}, trunk_flags);
+                layer.ffn_up_exps     = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS,   "weight", i), {moe_n_embd, n_ff_exp, n_expert}, trunk_flags);

                // Shared expert branch
-                layer.ffn_down_shexp  = create_tensor(tn(LLM_TENSOR_FFN_DOWN_SHEXP, "weight", i), {n_ff_shexp, n_embd}, 0);
-                layer.ffn_up_shexp    = create_tensor(tn(LLM_TENSOR_FFN_UP_SHEXP,   "weight", i), {n_embd, n_ff_shexp}, 0);
+                layer.ffn_down_shexp  = create_tensor(tn(LLM_TENSOR_FFN_DOWN_SHEXP, "weight", i), {n_ff_shexp, n_embd}, trunk_flags);
+                layer.ffn_up_shexp    = create_tensor(tn(LLM_TENSOR_FFN_UP_SHEXP,   "weight", i), {n_embd, n_ff_shexp}, trunk_flags);

            } else {
                // mlp layers
-                layer.ffn_down   = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {  hparams.n_ff(i), n_embd}, 0);
-                layer.ffn_up     = create_tensor(tn(LLM_TENSOR_FFN_UP,   "weight", i), {n_embd,   hparams.n_ff(i)}, 0);
+                layer.ffn_down   = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {  hparams.n_ff(i), n_embd}, trunk_flags);
+                layer.ffn_up     = create_tensor(tn(LLM_TENSOR_FFN_UP,   "weight", i), {n_embd,   hparams.n_ff(i)}, trunk_flags);
                layer.ffn_down_b = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "bias",   i), {n_embd}, TENSOR_NOT_REQUIRED);
                layer.ffn_up_b   = create_tensor(tn(LLM_TENSOR_FFN_UP,   "bias",   i), {hparams.n_ff(i)}, TENSOR_NOT_REQUIRED);
            }
        }
    }
+
+    // NextN/MTP draft head: each predict layer folds an attention sub-layer and a MoE
+    // sub-layer into a single trailing block
+    for (int i = n_layer; i < n_layer_all; ++i) {
+        auto & layer = layers[i];
+
+        const int64_t n_head_i       = hparams.n_head(i);
+        const int64_t n_embd_k_gqa_i = hparams.n_embd_k_gqa(i);
+        const int64_t n_embd_v_gqa_i = hparams.n_embd_v_gqa(i);
+        const int64_t n_ff_exp       = hparams.n_ff_exp ? hparams.n_ff_exp : n_ff / n_expert_used;
+        const int64_t n_ff_shexp     = hparams.n_ff_shexp;
+
+        // NextN input-fusion tensors
+        layer.nextn.enorm            = create_tensor(tn(LLM_TENSOR_NEXTN_ENORM,            "weight", i), {n_embd}, mtp_flags);
+        layer.nextn.hnorm            = create_tensor(tn(LLM_TENSOR_NEXTN_HNORM,            "weight", i), {n_embd}, mtp_flags);
+        layer.nextn.eh_proj          = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ,          "weight", i), {2*n_embd, n_embd}, mtp_flags);
+        layer.nextn.shared_head_norm = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_NORM, "weight", i), {n_embd}, mtp_flags);
+
+        // attention sub-layer
+        layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, mtp_flags);
+        create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head_i, n_embd_k_gqa_i, n_embd_v_gqa_i, mtp_flags);
+        layer.wo   = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_k * n_head_i, n_embd}, mtp_flags);
+        layer.wo_b = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "bias",   i), {n_embd}, mtp_flags | TENSOR_NOT_REQUIRED);
+
+        // MoE sub-layer
+        layer.attn_post_norm  = create_tensor(tn(LLM_TENSOR_ATTN_POST_NORM,  "weight", i), {n_embd}, mtp_flags);
+        layer.ffn_gate_inp    = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP,    "weight", i), {n_embd, n_expert}, mtp_flags);
+        layer.ffn_exp_probs_b = create_tensor(tn(LLM_TENSOR_FFN_EXP_PROBS_B, "bias",   i), {n_expert}, mtp_flags);
+        layer.ffn_down_exps   = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS,   "weight", i), {n_ff_exp,   moe_n_embd, n_expert}, mtp_flags);
+        layer.ffn_up_exps     = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS,     "weight", i), {moe_n_embd, n_ff_exp,   n_expert}, mtp_flags);
+        layer.ffn_down_shexp  = create_tensor(tn(LLM_TENSOR_FFN_DOWN_SHEXP,  "weight", i), {n_ff_shexp, n_embd}, mtp_flags);
+        layer.ffn_up_shexp    = create_tensor(tn(LLM_TENSOR_FFN_UP_SHEXP,    "weight", i), {n_embd, n_ff_shexp}, mtp_flags);
+    }
 }

 std::unique_ptr<llm_graph_context> llama_model_nemotron_h::build_arch_graph(const llm_graph_params & params) const {
@@ -153,7 +195,7 @@ llama_model_nemotron_h::graph::graph(const llama_model & model, const llm_graph_
            cur = build_ffn_layer(cur, model, il);
        }

-        if (il == n_layer - 1 && inp_out_ids) {
+        if (il == n_layer - 1 && inp_out_ids && cparams.embeddings_nextn_masked) {
            cur   = ggml_get_rows(ctx0, cur, inp_out_ids);
            inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
        }
@@ -170,6 +212,14 @@ llama_model_nemotron_h::graph::graph(const llama_model & model, const llm_graph_

    cur = build_norm(cur, model.output_norm, NULL, LLM_NORM_RMS, -1);

+    // seed for the MTP/NextN draft head
+    cb(cur, "h_nextn", -1);
+    res->t_h_nextn = cur;
+
+    if (!cparams.embeddings_nextn_masked && inp_out_ids) {
+        cur = ggml_get_rows(ctx0, cur, inp_out_ids);
+    }
+
    cb(cur, "result_norm", -1);
    res->t_embd = cur;

@@ -6712,19 +6712,26 @@ struct test_roll : public test_case {
    const int shift1;
    const int shift3;
    const int shift4;
+    const bool permute;

    std::string vars() override {
-        return VARS_TO_STR4(shift0, shift1, shift3, shift4);
+        return VARS_TO_STR5(shift0, shift1, shift3, shift4, permute);
    }

-    test_roll(int shift0 = 3, int shift1 = -2, int shift3 = 1, int shift4 = -1)
-        : shift0(shift0), shift1(shift1), shift3(shift3), shift4(shift4) {}
+    test_roll(int shift0 = 3, int shift1 = -2, int shift3 = 1, int shift4 = -1, bool permute = false)
+        : shift0(shift0), shift1(shift1), shift3(shift3), shift4(shift4), permute(permute) {}

    ggml_tensor * build_graph(ggml_context * ctx) override {
        int64_t ne[4] = {10, 5, 4, 3};
        ggml_tensor * a = ggml_new_tensor(ctx, GGML_TYPE_F32, 4, ne);
        ggml_set_name(a, "a");

+        if (permute) {
+            // ggml_roll only requires nb[0] == type size, so a permuted src is valid
+            a = ggml_permute(ctx, a, 0, 2, 1, 3);
+            ggml_set_name(a, "a_permuted");
+        }
+
        ggml_tensor * out = ggml_roll(ctx, a, shift0, shift1, shift3, shift4);
        ggml_set_name(out, "out");

@@ -9459,6 +9466,7 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
    test_cases.emplace_back(new test_pad_reflect_1d());
    test_cases.emplace_back(new test_pad_reflect_1d(GGML_TYPE_F32, {3000, 384, 4, 1}));
    test_cases.emplace_back(new test_roll());
+    test_cases.emplace_back(new test_roll(3, -2, 1, -1, true));
    test_cases.emplace_back(new test_arange());
    test_cases.emplace_back(new test_arange(GGML_TYPE_F32, 0.0f, 1048576.0f, 1.0f));
    test_cases.emplace_back(new test_timestep_embedding());
@@ -63,6 +63,7 @@ static void test_laguna_tool_format(testing & t);
 static void test_laguna_s_analysis(testing & t);
 static void test_laguna_s_reasoning_detection(testing & t);
 static void test_laguna_s_tool_format(testing & t);
+static void test_laguna_s_preserve_reasoning(testing & t);
 static void test_laguna_xs2_analysis(testing & t);
 static void test_laguna_xs2_reasoning_detection(testing & t);
 static void test_laguna_xs2_tool_format(testing & t);
@@ -1451,9 +1452,14 @@ static void test_laguna_s_tool_format(testing & t) {
    analysis.analyze_template(tmpl);
    t.assert_equal("Laguna-S(v8) arg_value_suffix should be '</arg_value>'", "</arg_value>", analysis.tools.arguments.value_suffix);
 }
+static void test_laguna_s_preserve_reasoning(testing & t) {
+    common_chat_template tmpl = load_laguna_s_template(t);
+    t.assert_true("Laguna-S(v8) supports preserving reasoning", tmpl.original_caps().supports_preserve_reasoning);
+}
 static void test_laguna_s_analysis(testing & t) {
    t.test("Laguna-S(v8) reasoning detection", test_laguna_s_reasoning_detection);
    t.test("Laguna-S(v8) tool format", test_laguna_s_tool_format);
+    t.test("Laguna-S(v8) preserve reasoning", test_laguna_s_preserve_reasoning);
 }

 static common_chat_template load_laguna_xs2_template(testing & t) {
@@ -192,7 +192,7 @@ static gguf_context_ptr get_gguf_ctx(const llm_arch arch, const bool moe) {
        ms.add_kv(LLM_KV_ROPE_FREQ_BASE_SWA,              10000.0f);
        // SWA pattern: every 5th layer is full attention (matches E2B layer_types)
        ms.add_kv(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, uint32_t(5));
-    } else if (arch == LLM_ARCH_COHERE2MOE || arch == LLM_ARCH_MIMO2 || arch == LLM_ARCH_STEP35) {
+    } else if (arch == LLM_ARCH_COHERE2MOE || arch == LLM_ARCH_MIMO2 || arch == LLM_ARCH_STEP35 || arch == LLM_ARCH_MUSE_GLIMMER) {
        std::vector<uint32_t> pattern;
        pattern.reserve(n_layer);
        for (uint32_t il = 0; il < n_layer; il++) {
@@ -54,6 +54,7 @@
 | `-ctv, --cache-type-v TYPE` | KV cache data type for V<br/>allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1<br/>(default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_V) |
 | `-dt, --defrag-thold N` | KV cache defragmentation threshold (DEPRECATED)<br/>(env: LLAMA_ARG_DEFRAG_THOLD) |
 | `-np, --parallel N` | number of parallel sequences to decode (default: 1)<br/>(env: LLAMA_ARG_N_PARALLEL) |
+| `--rpc SERVERS` | comma-separated list of RPC servers (host:port)<br/>(env: LLAMA_ARG_RPC) |
 | `--mlock` | DEPRECATED in favor of `--load-mode`: force system to keep model in RAM rather than swapping or compressing<br/>(env: LLAMA_ARG_MLOCK) |
 | `--mmap, --no-mmap` | DEPRECATED in favor of `--load-mode`: whether to memory-map model. (if mmap disabled, slower load but may reduce pageouts if not using mlock)<br/>(env: LLAMA_ARG_MMAP) |
 | `-dio, --direct-io, -ndio, --no-direct-io` | DEPRECATED in favor of `--load-mode`: use DirectIO if available<br/>(env: LLAMA_ARG_DIO) |
@@ -84,8 +85,6 @@
 | `-dr, --docker-repo [<repo>/]<model>[:quant]` | Docker Hub model repository. repo is optional, default to ai/. quant is optional, default to :latest.<br/>example: gemma3<br/>(default: unused)<br/>(env: LLAMA_ARG_DOCKER_REPO) |
 | `-hf, -hfr, --hf-repo <user>/<model>[:quant]` | Hugging Face model repository; quant is optional, case-insensitive, default to Q4_K_M, or falls back to the first file in the repo if Q4_K_M doesn't exist.<br/>mmproj is also downloaded automatically if available. to disable, add --no-mmproj<br/>example: ggml-org/GLM-4.7-Flash-GGUF:Q4_K_M<br/>(default: unused)<br/>(env: LLAMA_ARG_HF_REPO) |
 | `-hff, --hf-file FILE` | Hugging Face model file. If specified, it will override the quant in --hf-repo (default: unused)<br/>(env: LLAMA_ARG_HF_FILE) |
-| `-hfv, -hfrv, --hf-repo-v <user>/<model>[:quant]` | Hugging Face model repository for the vocoder model (default: unused)<br/>(env: LLAMA_ARG_HF_REPO_V) |
-| `-hffv, --hf-file-v FILE` | Hugging Face model file for the vocoder model (default: unused)<br/>(env: LLAMA_ARG_HF_FILE_V) |
 | `-hft, --hf-token TOKEN` | Hugging Face access token (default: value from HF_TOKEN environment variable)<br/>(env: HF_TOKEN) |
 | `--log-disable` | Log disable |
 | `--log-file FNAME` | Log to file<br/>(env: LLAMA_ARG_LOG_FILE) |
@@ -137,6 +137,7 @@ llama-completion.exe -m models\gemma-1.1-7b-it.Q4_K_M.gguf --ignore-eos -n -1
 | `-ctv, --cache-type-v TYPE` | KV cache data type for V<br/>allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1<br/>(default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_V) |
 | `-dt, --defrag-thold N` | KV cache defragmentation threshold (DEPRECATED)<br/>(env: LLAMA_ARG_DEFRAG_THOLD) |
 | `-np, --parallel N` | number of parallel sequences to decode (default: 1)<br/>(env: LLAMA_ARG_N_PARALLEL) |
+| `--rpc SERVERS` | comma-separated list of RPC servers (host:port)<br/>(env: LLAMA_ARG_RPC) |
 | `--mlock` | DEPRECATED in favor of `--load-mode`: force system to keep model in RAM rather than swapping or compressing<br/>(env: LLAMA_ARG_MLOCK) |
 | `--mmap, --no-mmap` | DEPRECATED in favor of `--load-mode`: whether to memory-map model. (if mmap disabled, slower load but may reduce pageouts if not using mlock)<br/>(env: LLAMA_ARG_MMAP) |
 | `-dio, --direct-io, -ndio, --no-direct-io` | DEPRECATED in favor of `--load-mode`: use DirectIO if available<br/>(env: LLAMA_ARG_DIO) |
@@ -167,8 +168,6 @@ llama-completion.exe -m models\gemma-1.1-7b-it.Q4_K_M.gguf --ignore-eos -n -1
 | `-dr, --docker-repo [<repo>/]<model>[:quant]` | Docker Hub model repository. repo is optional, default to ai/. quant is optional, default to :latest.<br/>example: gemma3<br/>(default: unused)<br/>(env: LLAMA_ARG_DOCKER_REPO) |
 | `-hf, -hfr, --hf-repo <user>/<model>[:quant]` | Hugging Face model repository; quant is optional, case-insensitive, default to Q4_K_M, or falls back to the first file in the repo if Q4_K_M doesn't exist.<br/>mmproj is also downloaded automatically if available. to disable, add --no-mmproj<br/>example: ggml-org/GLM-4.7-Flash-GGUF:Q4_K_M<br/>(default: unused)<br/>(env: LLAMA_ARG_HF_REPO) |
 | `-hff, --hf-file FILE` | Hugging Face model file. If specified, it will override the quant in --hf-repo (default: unused)<br/>(env: LLAMA_ARG_HF_FILE) |
-| `-hfv, -hfrv, --hf-repo-v <user>/<model>[:quant]` | Hugging Face model repository for the vocoder model (default: unused)<br/>(env: LLAMA_ARG_HF_REPO_V) |
-| `-hffv, --hf-file-v FILE` | Hugging Face model file for the vocoder model (default: unused)<br/>(env: LLAMA_ARG_HF_FILE_V) |
 | `-hft, --hf-token TOKEN` | Hugging Face access token (default: value from HF_TOKEN environment variable)<br/>(env: HF_TOKEN) |
 | `--log-disable` | Log disable |
 | `--log-file FNAME` | Log to file<br/>(env: LLAMA_ARG_LOG_FILE) |
@@ -43,6 +43,7 @@ add_library(mtmd
            models/kimivl.cpp
            models/kimik25.cpp
            models/nemotron-v2-vl.cpp
+            models/muse-glimmer.cpp
            models/llama4.cpp
            models/llava.cpp
            models/minicpmv.cpp
@@ -455,6 +455,7 @@ enum projector_type {
    PROJECTOR_TYPE_MIMO_AUDIO,
    PROJECTOR_TYPE_QWEN3TTS_SPKENC,
    PROJECTOR_TYPE_QWEN3TTS_GEN,
+    PROJECTOR_TYPE_MUSE_GLIMMER,
    PROJECTOR_TYPE_UNKNOWN,
 };

@@ -514,6 +515,7 @@ static std::map<projector_type, std::string> PROJECTOR_TYPE_NAMES = {
    { PROJECTOR_TYPE_PARAKEET,          "parakeet"},
    { PROJECTOR_TYPE_QWEN3TTS_SPKENC,   "qwen3tts_spkenc"},
    { PROJECTOR_TYPE_QWEN3TTS_GEN,      "qwen3tts_gen"},
+    { PROJECTOR_TYPE_MUSE_GLIMMER,      "muse-glimmer"},
 };

 static projector_type clip_projector_type_from_string(const std::string & str) {
@@ -109,6 +109,11 @@ struct clip_hparams {
    int32_t downsample_query_side;
    int32_t downsample_window_side;

+    // Muse Glimmer vision (per-block sparse-window pattern, learned pos-emb, patch-temporal)
+    // NOTE: these perhaps shouldn't have the architecture prefix
+    int32_t muse_glimmer_patch_temporal = 0;
+    int32_t muse_glimmer_sparse_factor  = 0;
+
    // audio
    int32_t n_mel_bins = 0; // whisper preprocessor
    int32_t proj_stack_factor = 0; // ultravox
@@ -954,6 +954,10 @@ static std::unique_ptr<clip_graph> clip_get_graph_builder(clip_ctx * ctx, const
            {
                builder = std::make_unique<clip_graph_minimax_m3>(ctx, img);
            } break;
+        case PROJECTOR_TYPE_MUSE_GLIMMER:
+            {
+                builder = std::make_unique<clip_graph_muse_glimmer>(ctx, img);
+            } break;
        case PROJECTOR_TYPE_STEP3VL:
            {
                builder = std::make_unique<clip_graph_step3vl>(ctx, img);
@@ -1572,6 +1576,17 @@ struct clip_model_loader {
                        hparams.set_limit_image_tokens(8, 576);
                        hparams.set_warmup_n_tokens(16*16);
                    } break;
+                case PROJECTOR_TYPE_MUSE_GLIMMER:
+                    {
+                        hparams.n_merge = 2; // pixel-shuffle downsample after the ViT
+                        hparams.image_resize_algo = RESIZE_ALGO_LANCZOS;
+                        hparams.rope_theta = 10000.0f;
+                        hparams.muse_glimmer_patch_temporal = 2;
+                        hparams.muse_glimmer_sparse_factor  = 4; // 3 sparse layers + 1 global, repeating
+                        get_u32(KEY_SPATIAL_MERGE_SIZE,  hparams.n_merge,             false);
+                        hparams.set_limit_image_tokens(1, 4096);
+                        hparams.set_warmup_n_tokens(32*32);
+                    } break;
                case PROJECTOR_TYPE_MIMOVL:
                    {
                        hparams.n_merge = 2; // spatial_merge_size
@@ -2317,6 +2332,13 @@ struct clip_model_loader {
                    model.mm_merger_fc2_w = get_tensor(string_format(TN_MM_MERGER_FC2, "weight"));
                    model.mm_merger_fc2_b = get_tensor(string_format(TN_MM_MERGER_FC2, "bias"));
                } break;
+            case PROJECTOR_TYPE_MUSE_GLIMMER:
+                {
+                    // 3-linear MLP: fc -> erf-GELU -> proj -> erf-GELU -> vision_proj (into LLM residual dim)
+                    model.mm_0_w = get_tensor(string_format(TN_LLAVA_PROJ, 0, "weight"));
+                    model.mm_1_w = get_tensor(string_format(TN_LLAVA_PROJ, 1, "weight"));
+                    model.mm_2_w = get_tensor(string_format(TN_LLAVA_PROJ, 2, "weight"));
+                } break;
            case PROJECTOR_TYPE_STEP3VL:
                {
                    model.mm_0_w     = get_tensor(string_format(TN_LLAVA_PROJ, 0, "weight"));
@@ -3745,6 +3767,7 @@ int clip_n_output_tokens_x(const clip_ctx * ctx, const clip_image_f32 * img) {
        case PROJECTOR_TYPE_PADDLEOCR:
        case PROJECTOR_TYPE_HUNYUANVL:
        case PROJECTOR_TYPE_YOUTUVL:
+        case PROJECTOR_TYPE_MUSE_GLIMMER:
            return (img->nx() / params.patch_size) / 2;
        case PROJECTOR_TYPE_STEP3VL:
            return img->nx() / (params.patch_size * params.n_merge);
@@ -3770,6 +3793,7 @@ int clip_n_output_tokens_y(const clip_ctx * ctx, const clip_image_f32 * img) {
        case PROJECTOR_TYPE_PADDLEOCR:
        case PROJECTOR_TYPE_HUNYUANVL:
        case PROJECTOR_TYPE_YOUTUVL:
+        case PROJECTOR_TYPE_MUSE_GLIMMER:
            return (img->ny() / params.patch_size) / 2;
        case PROJECTOR_TYPE_STEP3VL:
            return img->ny() / (params.patch_size * params.n_merge);
@@ -3848,6 +3872,7 @@ int clip_n_output_tokens(const clip_ctx * ctx, const clip_image_f32 * img) {
        case PROJECTOR_TYPE_MINIMAX_M3:
        case PROJECTOR_TYPE_GLM4V:
        case PROJECTOR_TYPE_YOUTUVL:
+        case PROJECTOR_TYPE_MUSE_GLIMMER:
            {
                // dynamic size (2 conv, so double patch size)
                int x_patch = img->nx() / (params.patch_size * 2);
@@ -4193,6 +4218,70 @@ bool clip_encode(struct clip_ctx * ctx, struct clip_encode_params * params) {

    // set input per projector
    switch (ctx->model.proj_type) {
+        case PROJECTOR_TYPE_MUSE_GLIMMER:
+            {
+                const int grid_w = pos_w;            // image_size_width  / patch_size
+                const int grid_h = pos_h;            // image_size_height / patch_size
+                const int n_tok  = grid_w * grid_h;
+                const int pgrid  = (int) std::sqrt((double) ctx->model.position_embeddings->ne[1]); // 32
+                const int f      = hparams.n_merge;  // downsample 2
+
+                // pixel patchify runs inside the graph via build_inp() (ggml_conv_2d);
+                // pos-emb bilinear interp via resize_position_embeddings().
+
+                // --- sparse window grouping (pgrid x pgrid windows) ---
+                const int win = pgrid;
+                const int nwin_h = (grid_h + win - 1) / win;
+                const int nwin_w = (grid_w + win - 1) / win;
+                std::vector<int32_t> sp_perm; sp_perm.reserve(n_tok);
+                std::vector<int>     sp_slens;
+                for (int wy = 0; wy < nwin_h; wy++) {
+                    for (int wx = 0; wx < nwin_w; wx++) {
+                        int cnt = 0;
+                        for (int hh = 0; hh < win; hh++) {
+                            for (int ww = 0; ww < win; ww++) {
+                                const int gy = wy * win + hh;
+                                const int gx = wx * win + ww;
+                                if (gy < grid_h && gx < grid_w) { sp_perm.push_back(gy * grid_w + gx); cnt++; }
+                            }
+                        }
+                        if (cnt > 0) sp_slens.push_back(cnt);
+                    }
+                }
+                std::vector<int32_t> rpos_w(n_tok), rpos_h(n_tok), inv_perm(n_tok);
+                for (int i = 0; i < n_tok; i++) {
+                    const int orig = sp_perm[i];
+                    rpos_w[i] = (orig % grid_w) + 1; // 1-indexed
+                    rpos_h[i] = (orig / grid_w) + 1;
+                    inv_perm[orig] = i;
+                }
+                set_input_i32("muse_glimmer_sp_perm",  sp_perm);
+                set_input_i32("muse_glimmer_inv_perm", inv_perm);
+                set_input_i32("muse_glimmer_pos_w",    rpos_w);
+                set_input_i32("muse_glimmer_pos_h",    rpos_h);
+
+                // block-diagonal window mask (permuted order)
+                std::vector<float> sp_mask((size_t) n_tok * n_tok, -INFINITY);
+                {
+                    int off = 0;
+                    for (int s : sp_slens) {
+                        for (int a = 0; a < s; a++)
+                            for (int b = 0; b < s; b++)
+                                sp_mask[(size_t) (off + a) * n_tok + (off + b)] = 0.0f;
+                        off += s;
+                    }
+                }
+                set_input_f32("muse_glimmer_sp_mask", sp_mask);
+
+                // pixel-shuffle gather (original order): f*f spatial neighbours grouped
+                std::vector<int32_t> dsp; dsp.reserve(n_tok);
+                for (int oy = 0; oy < grid_h / f; oy++)
+                    for (int ox = 0; ox < grid_w / f; ox++)
+                        for (int ry = 0; ry < f; ry++)
+                            for (int rx = 0; rx < f; rx++)
+                                dsp.push_back((oy * f + ry) * grid_w + (ox * f + rx));
+                set_input_i32("muse_glimmer_ds_perm", dsp);
+            } break;
        case PROJECTOR_TYPE_MINICPMV:
            {
                // inspired from siglip:
@@ -5369,6 +5458,8 @@ int clip_n_mmproj_embd(const struct clip_ctx * ctx) {
            return ctx->model.mm_model_mlp_3_w->ne[1];
        case PROJECTOR_TYPE_MINIMAX_M3:
            return ctx->model.mm_merger_fc2_b->ne[0];
+        case PROJECTOR_TYPE_MUSE_GLIMMER:
+            return ctx->model.mm_2_w->ne[1];
        case PROJECTOR_TYPE_QWEN2VL:
        case PROJECTOR_TYPE_QWEN25VL:
        case PROJECTOR_TYPE_EXAONE4_5:
@@ -365,3 +365,8 @@ private:
    ggml_tensor * build_newline_row(ggml_context * ctx0);
    ggml_tensor * append_rowwise_newlines(ggml_context * ctx0, ggml_tensor * tile_output);
 };
+
+struct clip_graph_muse_glimmer : clip_graph {
+    clip_graph_muse_glimmer(clip_ctx * ctx, const clip_image_f32 & img) : clip_graph(ctx, img) {}
+    ggml_cgraph * build() override;
+};
@@ -0,0 +1,88 @@
+#include "models.h"
+
+// MuseGlimmer vision encoder: 50-layer ViT with 2D RoPE, sparse block-diagonal
+// window attention (every 4th + last layer global), pixel-shuffle downsample, then
+// adapter MLP + LLM's vision_projection.
+//
+// Several quantities are precomputed on host and fed as named graph inputs (filled in
+// clip.cpp set_input, PROJECTOR_TYPE_MUSE_GLIMMER branch):
+//   muse_glimmer_pos_w/_h [n_tok] i32         : 1-indexed RoPE positions (sparse-permuted order)
+//   muse_glimmer_sp_perm  [n_tok] i32         : window grouping permutation (applied after ln_pre)
+//   muse_glimmer_inv_perm [n_tok] i32         : inverse of sp_perm (applied after blocks)
+//   muse_glimmer_ds_perm  [n_tok] i32         : pixel-shuffle gather (original order)
+//   muse_glimmer_sp_mask  [n_tok, n_tok] f32  : block-diagonal window mask (sparse layers)
+ggml_cgraph * clip_graph_muse_glimmer::build() {
+    const int ds = hparams.n_merge;              // downsample factor (2)
+    const int sf = hparams.muse_glimmer_sparse_factor;   // 4
+    const int n_tok     = n_patches;
+    const int n_out     = (n_patches_x / ds) * (n_patches_y / ds);
+    const float rope_base = hparams.rope_theta;  // 10000
+
+    auto inp_i32 = [&](const char * name, int64_t n) {
+        ggml_tensor * t = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n);
+        ggml_set_name(t, name);
+        ggml_set_input(t);
+        return t;
+    };
+
+    ggml_tensor * pos_w    = inp_i32("muse_glimmer_pos_w",    n_tok);
+    ggml_tensor * pos_h    = inp_i32("muse_glimmer_pos_h",    n_tok);
+    ggml_tensor * sp_perm  = inp_i32("muse_glimmer_sp_perm",  n_tok);
+    ggml_tensor * inv_perm = inp_i32("muse_glimmer_inv_perm", n_tok);
+    ggml_tensor * ds_perm  = inp_i32("muse_glimmer_ds_perm",  n_tok);
+
+    ggml_tensor * sp_mask = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_tok, n_tok);
+    ggml_set_name(sp_mask, "muse_glimmer_sp_mask");
+    ggml_set_input(sp_mask);
+
+    // patchify via build_inp (conv2d over raw pixels) + bilinear-resized learned pos-emb
+    ggml_tensor * x = build_inp();                                                     // [n_embd, n_tok, 1]
+    x = ggml_add(ctx0, x, resize_position_embeddings(GGML_SCALE_MODE_BILINEAR));
+    cb(x, "after_posemb", -1);
+
+    // group patches into pgrid x pgrid windows (sparse attention order)
+    x = ggml_get_rows(ctx0, x, sp_perm);
+    cb(x, "after_sp_perm", -1);
+
+    // per-layer mask: sparse layers get sp_mask, global layers (every sf-th and last) get none
+    std::vector<ggml_tensor *> attn_mask_layers(n_layer);
+    for (int il = 0; il < n_layer; ++il) {
+        const bool is_global = (il == n_layer - 1) || ((il + 1) % sf == 0);
+        attn_mask_layers[il] = is_global ? nullptr : sp_mask;
+    }
+
+    // 2D RoPE: first half of head_dim uses width pos, second half uses height pos
+    auto add_pos = [&](ggml_tensor * cur, const clip_layer &) {
+        return build_rope_2d(ctx0, cur, pos_w, pos_h, rope_base, false);
+    };
+
+    build_vit_opts opts;
+    opts.attn_mask_layers = std::move(attn_mask_layers);
+
+    // pre_ln, per-layer transformer, post_ln (all inside build_vit); reference uses exact (erf) GELU
+    x = build_vit(x, n_tok, NORM_TYPE_NORMAL, FFN_GELU_ERF, nullptr, add_pos, opts);
+
+    // un-permute back to original grid order
+    x = ggml_get_rows(ctx0, x, inv_perm);
+    cb(x, "after_inv_perm", -1);
+
+    // pixel-shuffle downsample: gather f*f spatial neighbors then concat channel-outer.
+    // out[c*(ds*ds)+s, o] = x[ds_perm gathered][o*(ds*ds)+s, c]
+    x = ggml_get_rows(ctx0, x, ds_perm);                 // [n_embd, n_tok], grouped
+    x = ggml_reshape_3d(ctx0, x, n_embd, ds * ds, n_out);// [c, s, o]
+    x = ggml_permute(ctx0, x, 1, 0, 2, 3);               // [s, c, o]
+    x = ggml_cont(ctx0, x);
+    x = ggml_reshape_2d(ctx0, x, n_embd * ds * ds, n_out); // [6144, n_out]
+    cb(x, "encoder_out", -1);
+
+    // adapter (6144->4096->4096, exact GELU each) + LLM vision_projection (4096->6656)
+    x = build_mm(model.mm_0_w, x);
+    x = ggml_gelu_erf(ctx0, x);
+    x = build_mm(model.mm_1_w, x);
+    x = ggml_gelu_erf(ctx0, x);
+    x = build_mm(model.mm_2_w, x);                       // [6656, n_out]
+    cb(x, "projected", -1);
+
+    ggml_build_forward_expand(gf, x);
+    return gf;
+}
@@ -1615,3 +1615,65 @@ mtmd_image_preproc_out mtmd_image_preprocessor_granite::preprocess(const clip_im
    }
    return output;
 }
+
+//
+// mtmd_image_preprocessor_muse_glimmer
+//
+
+// Replicates transformers' get_aspect_ratio_preserving_size
+static clip_image_size muse_glimmer_grid_size(int img_w, int img_h, int patch_hw, int max_tokens) {
+    double i_nph = (double) img_h / patch_hw;
+    double i_npw = (double) img_w / patch_hw;
+    const double ratio = i_nph > 0.0 ? i_npw / i_nph : 1.0;
+    if (i_nph * i_npw > (double) max_tokens) {
+        i_nph = std::sqrt((double) max_tokens / ratio);
+        i_npw = i_nph * ratio;
+    }
+    const int hs[2] = { (int) std::floor(i_nph), (int) std::ceil(i_nph) };
+    const int ws[2] = { (int) std::floor(i_npw), (int) std::ceil(i_npw) };
+    const double target_ar = (double) img_h / (double) img_w;
+    int    best_nph = -1;
+    int    best_npw = -1;
+    double best_d   = 0.0;
+    for (int a = 0; a < 2; ++a) {
+        for (int b = 0; b < 2; ++b) {
+            const int nph = hs[a];
+            const int npw = ws[b];
+            if (nph < 1 || npw < 1 || nph * npw > max_tokens) {
+                continue;
+            }
+            const double d = std::fabs((double) nph / (double) npw - target_ar);
+            const int n_tokens      = nph * npw;
+            const int best_n_tokens = best_nph * best_npw;
+            if (best_nph < 0 || d < best_d || (d == best_d && n_tokens > best_n_tokens)) {
+                best_nph = nph;
+                best_npw = npw;
+                best_d   = d;
+            }
+        }
+    }
+    if (best_nph < 0) { // no candidate fit under the cap: round and clamp
+        best_nph = std::max(1, (int) std::lround(i_nph));
+        best_npw = std::max(1, (int) std::lround(i_npw));
+    }
+    return clip_image_size{ best_npw * patch_hw, best_nph * patch_hw };
+}
+
+mtmd_image_preproc_out mtmd_image_preprocessor_muse_glimmer::preprocess(const clip_image_u8 & img) {
+    const int patch_hw   = hparams.patch_size * hparams.n_merge;
+    const int patch_area = hparams.patch_size * hparams.patch_size * hparams.n_merge * hparams.n_merge;
+    GGML_ASSERT(patch_area > 0 && hparams.image_max_pixels > 0);
+    const int max_tokens = hparams.image_max_pixels / patch_area;
+
+    const clip_image_size original_size = img.get_size();
+    const clip_image_size target_size   = muse_glimmer_grid_size(
+        original_size.width, original_size.height, patch_hw, max_tokens);
+
+    // PIL resizes directly to (target_w, target_h) -- a stretch, no padding.
+    clip_image_u8 resized_image;
+    img_tool::resize(img, resized_image, target_size, hparams.image_resize_algo, PAD_NONE);
+
+    mtmd_image_preproc_out output;
+    output.append(hparams, resized_image, true);
+    return output;
+}
@@ -230,3 +230,9 @@ struct mtmd_image_preprocessor_granite : mtmd_image_preprocessor_llava_uhd {
    mtmd_image_preprocessor_granite(const clip_ctx * ctx) : mtmd_image_preprocessor_llava_uhd(ctx) {}
    mtmd_image_preproc_out preprocess(const clip_image_u8 & img) override;
 };
+
+// pick the patch grid closest to the input aspect ratio under the per-image token cap, stretch-resize.
+struct mtmd_image_preprocessor_muse_glimmer : mtmd_image_preprocessor {
+    mtmd_image_preprocessor_muse_glimmer(const clip_ctx * ctx) : mtmd_image_preprocessor(ctx) {}
+    mtmd_image_preproc_out preprocess(const clip_image_u8 & img) override;
+};
@@ -699,6 +699,12 @@ struct mtmd_context {
                    img_end = "]<]end of image[>[";
                    image_preproc = std::make_unique<mtmd_image_preprocessor_dyn_size>(ctx_v);
                } break;
+            case PROJECTOR_TYPE_MUSE_GLIMMER:
+                {
+                    img_beg = "<|image_start|>";
+                    img_end = "<|image_end|>";
+                    image_preproc = std::make_unique<mtmd_image_preprocessor_muse_glimmer>(ctx_v);
+                } break;
            case PROJECTOR_TYPE_YOUTUVL:
                {
                    // <|vision_start|> ... (image embeddings) ... <|vision_end|>
@@ -201,7 +201,7 @@ Invoke a tool call, request body is a JSON object with:

 Headers:
 - `x-tool-cwd`: optional; if set, use as the CWD for tool; this is not part of tool's params because it's meant to be set by the runtime, not the LLM itself
- `x-tool-runtime`: optional; if set, run the tool inside this isolate instead of on the host. Only `docker-container:<id>` is supported for now, using an already-running container
+- `x-tool-runtime`: optional; if set, run the tool inside this isolate instead of on the host. Either `docker-container:<id>` or `podman-container:<id>`, using an already-running container, or `ssh:<target>`, running the tool on a remote host

 Returns JSON object. There are two response formats (MCP tools use the same two formats: their result content is concatenated into `plain_text_response`, and RPC or tool errors are surfaced as the `error` string):

@@ -71,6 +71,7 @@ For the full list of features, please refer to [server's changelog](https://gith
 | `-ctk, --cache-type-k TYPE` | KV cache data type for K<br/>allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1<br/>(default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_K) |
 | `-ctv, --cache-type-v TYPE` | KV cache data type for V<br/>allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1<br/>(default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_V) |
 | `-dt, --defrag-thold N` | KV cache defragmentation threshold (DEPRECATED)<br/>(env: LLAMA_ARG_DEFRAG_THOLD) |
+| `--rpc SERVERS` | comma-separated list of RPC servers (host:port)<br/>(env: LLAMA_ARG_RPC) |
 | `--mlock` | DEPRECATED in favor of `--load-mode`: force system to keep model in RAM rather than swapping or compressing<br/>(env: LLAMA_ARG_MLOCK) |
 | `--mmap, --no-mmap` | DEPRECATED in favor of `--load-mode`: whether to memory-map model. (if mmap disabled, slower load but may reduce pageouts if not using mlock)<br/>(env: LLAMA_ARG_MMAP) |
 | `-dio, --direct-io, -ndio, --no-direct-io` | DEPRECATED in favor of `--load-mode`: use DirectIO if available<br/>(env: LLAMA_ARG_DIO) |
@@ -101,8 +102,6 @@ For the full list of features, please refer to [server's changelog](https://gith
 | `-dr, --docker-repo [<repo>/]<model>[:quant]` | Docker Hub model repository. repo is optional, default to ai/. quant is optional, default to :latest.<br/>example: gemma3<br/>(default: unused)<br/>(env: LLAMA_ARG_DOCKER_REPO) |
 | `-hf, -hfr, --hf-repo <user>/<model>[:quant]` | Hugging Face model repository; quant is optional, case-insensitive, default to Q4_K_M, or falls back to the first file in the repo if Q4_K_M doesn't exist.<br/>mmproj is also downloaded automatically if available. to disable, add --no-mmproj<br/>example: ggml-org/GLM-4.7-Flash-GGUF:Q4_K_M<br/>(default: unused)<br/>(env: LLAMA_ARG_HF_REPO) |
 | `-hff, --hf-file FILE` | Hugging Face model file. If specified, it will override the quant in --hf-repo (default: unused)<br/>(env: LLAMA_ARG_HF_FILE) |
-| `-hfv, -hfrv, --hf-repo-v <user>/<model>[:quant]` | Hugging Face model repository for the vocoder model (default: unused)<br/>(env: LLAMA_ARG_HF_REPO_V) |
-| `-hffv, --hf-file-v FILE` | Hugging Face model file for the vocoder model (default: unused)<br/>(env: LLAMA_ARG_HF_FILE_V) |
 | `-hft, --hf-token TOKEN` | Hugging Face access token (default: value from HF_TOKEN environment variable)<br/>(env: HF_TOKEN) |
 | `--log-disable` | Log disable |
 | `--log-file FNAME` | Log to file<br/>(env: LLAMA_ARG_LOG_FILE) |
@@ -197,9 +196,8 @@ For the full list of features, please refer to [server's changelog](https://gith
 | `--ui-config, --webui-config JSON` | JSON that provides default UI settings (overrides UI defaults)<br/>(env: LLAMA_ARG_UI_CONFIG) |
 | `--ui-config-file, --webui-config-file PATH` | JSON file that provides default UI settings (overrides UI defaults)<br/>(env: LLAMA_ARG_UI_CONFIG_FILE) |
 | `--ui-mcp-proxy, --webui-mcp-proxy, --no-ui-mcp-proxy, --no-webui-mcp-proxy` | experimental: whether to enable MCP CORS proxy - do not enable in untrusted environments (default: disabled)<br/>(env: LLAMA_ARG_UI_MCP_PROXY) |
-| `--tools TOOL1,TOOL2,...` | experimental: whether to enable built-in tools for AI agents - do not enable in untrusted environments (default: no tools)<br/>specify "all" to enable all tools<br/>available tools: read_file, file_glob_search, grep_search, exec_shell_command, write_file, edit_file, get_datetime<br/>note: for security reasons, this will limit --cors-origins to localhost by default<br/>(env: LLAMA_ARG_TOOLS) |
-| `--tools-runtime OPTION` | experimental: run tools in a separate runtime environment (default: none, use host environment)<br/>available options:<br/>  'docker:<image>': spin up a new Docker container and reuse it for all invocations, clean up on server exit<br/>  'docker-container:<id>': use an existing Docker container by ID, won't stop on server exit<br/><br/>(env: LLAMA_ARG_TOOLS_RUNTIME) |
 | `--tools TOOL1,TOOL2,...` | experimental: whether to enable built-in tools for AI agents - do not enable in untrusted environments (default: no tools)<br/>specify "all" to enable all tools<br/>available tools: read_file, file_glob_search, grep_search, exec_shell_command, write_file, edit_file, get_datetime, get_info<br/>note: for security reasons, this will limit --cors-origins to localhost by default<br/>(env: LLAMA_ARG_TOOLS) |
+| `--tools-runtime OPTION` | experimental: run tools in a separate runtime environment (default: none, use host environment)<br/>available options:<br/>  'docker:<image>', 'podman:<image>': spin up a new container and reuse it for all invocations, clean up on server exit<br/>  'docker-container:<id>', 'podman-container:<id>': use an existing container by ID, won't stop on server exit<br/>  'ssh:<target>': run tools on a remote POSIX host over SSH, key-based auth and a trusted host key are required<br/><br/>(env: LLAMA_ARG_TOOLS_RUNTIME) |
 | `--mcp-servers-config PATH` | experimental: path to JSON file with MCP server definitions (Cursor-compatible format) - do not enable in untrusted environments (default: none)<br/>note: for security reasons, this will limit --cors-origins to localhost by default<br/>(env: LLAMA_ARG_MCP_SERVERS_CONFIG) |
 | `--mcp-servers-json JSON` | experimental: inline JSON with MCP server definitions (Cursor-compatible format) - do not enable in untrusted environments (default: none)<br/>note: for security reasons, this will limit --cors-origins to localhost by default<br/>(env: LLAMA_ARG_MCP_SERVERS_JSON) |
 | `-ag, --agent, -no-ag, --no-agent` | whether to enable CORS proxy and all built-in tools - do not enable in untrusted environments (default: disabled)<br/>note: for security reasons, this will limit --cors-origins to localhost by default<br/>(env: LLAMA_ARG_AGENT) |
@@ -280,8 +278,6 @@ For the full list of features, please refer to [server's changelog](https://gith
 | `--spec-ngram-size-n N` | the argument has been removed. use the respective --spec-ngram-*-size-n or --spec-ngram-mod-n-match |
 | `--spec-ngram-size-m N` | the argument has been removed. use the respective --spec-ngram-*-size-m |
 | `--spec-ngram-min-hits N` | the argument has been removed. use the respective --spec-ngram-*-min-hits |
-| `-mv, --model-vocoder FNAME` | vocoder model for audio generation (default: unused) |
-| `--tts-use-guide-tokens` | Use guide tokens to improve TTS word recall |
 | `--embd-gemma-default` | use default EmbeddingGemma model (note: can download weights from the internet) |
 | `--fim-qwen-1.5b-default` | use default Qwen 2.5 Coder 1.5B (note: can download weights from the internet) |
 | `--fim-qwen-3b-default` | use default Qwen 2.5 Coder 3B (note: can download weights from the internet) |
@@ -10,6 +10,7 @@
 #include <ctime>
 #include <atomic>
 #include <cstring>
+#include <cctype>
 #include <cstdint>
 #include <cstdlib>
 #include <algorithm>
@@ -25,6 +26,11 @@
 #       define NOMINMAX
 #   endif
 #   include <windows.h>
+#   include <fcntl.h>
+#   include <io.h>
+#else
+#   include <cerrno>
+#   include <unistd.h>
 #endif

 namespace fs = std::filesystem;
@@ -176,7 +182,7 @@ public:
            const std::function<bool(const std::string &)> & on_chunk = nullptr) const = 0;
 };

-// shared subprocess execution helper, used by both the local and the docker-backed tools_io implementations.
+// shared subprocess execution helper, used by both the local and the isolate-backed tools_io implementations.
 // combine_stderr=false when the raw stdout bytes must not be tainted by stderr, e.g. reading file contents.
 static tools_io::exec_result run_subprocess(
        const std::vector<std::string> & args,
@@ -184,7 +190,8 @@ static tools_io::exec_result run_subprocess(
        int timeout_secs,
        const std::function<bool(const std::string &)> & on_chunk,
        bool combine_stderr,
-        const std::string & cwd = "") {
+        const std::string & cwd = "",
+        const std::string * stdin_data = nullptr) {
    tools_io::exec_result res;

    common_subproc proc;
@@ -216,26 +223,59 @@ static tools_io::exec_result run_subprocess(
        }
    });

+    // write stdin before reading stdout, the child drains stdin as it goes
+    // always close stdin, a transport client waits forever if its stdin pipe stays open
+    if (FILE * in = proc.stdin_file()) {
+        if (stdin_data != nullptr && !stdin_data->empty()) {
+#if defined(_WIN32)
+            // pipe fds default to CRT text mode: binary keeps the bytes untranslated
+            _setmode(_fileno(in), _O_BINARY);
+#endif
+            // a short write is not an error by itself, the exit code below decides
+            fwrite(stdin_data->data(), 1, stdin_data->size(), in);
+        }
+        fflush(in);
+    }
+    proc.close_stdin();
+
    FILE * f = proc.stdout_file();
    std::string output;
    bool truncated = false;
    if (f) {
+#if defined(_WIN32)
+        // pipe fds default to CRT text mode: binary keeps the bytes untranslated
+        _setmode(_fileno(f), _O_BINARY);
+#endif
+        // read raw bytes, not lines: the output can hold NUL and must arrive as soon as it is ready
+        // keep draining past the size cap, else the child blocks on a full pipe
        char buf[4096];
-        while (fgets(buf, sizeof(buf), f) != nullptr) {
-            if (!truncated) {
-                size_t len = strlen(buf);
-                if (output.size() + len <= max_output) {
-                    output.append(buf, len);
-                    if (on_chunk && !on_chunk(console_output_to_utf8(std::string(buf, len)))) {
-                        proc.terminate();
-                        break;
-                    }
-                } else {
-                    size_t remaining = max_output - output.size();
-                    output.append(buf, remaining);
-                    if (on_chunk && remaining > 0) on_chunk(console_output_to_utf8(std::string(buf, remaining)));
-                    truncated = true;
+        for (;;) {
+#if defined(_WIN32)
+            const int n = _read(_fileno(f), buf, (unsigned) sizeof(buf));
+#else
+            ssize_t n = read(fileno(f), buf, sizeof(buf));
+            while (n < 0 && errno == EINTR) {
+                n = read(fileno(f), buf, sizeof(buf));
+            }
+#endif
+            if (n <= 0) {
+                break;
+            }
+            if (truncated) {
+                continue;
+            }
+            const size_t len = (size_t) n;
+            if (output.size() + len <= max_output) {
+                output.append(buf, len);
+                if (on_chunk && !on_chunk(console_output_to_utf8(std::string(buf, len)))) {
+                    proc.terminate();
+                    break;
                }
+            } else {
+                size_t remaining = max_output - output.size();
+                output.append(buf, remaining);
+                if (on_chunk && remaining > 0) on_chunk(console_output_to_utf8(std::string(buf, remaining)));
+                truncated = true;
            }
        }
    }
@@ -473,7 +513,7 @@ private:
    }
 };

-// timeout for auxiliary isolate calls (stat/mkdir/ls/cp helpers); exec_shell_command uses its own
+// timeout for auxiliary isolate calls (stat/mkdir/ls helpers); exec_shell_command uses its own
 // caller-controlled timeout instead, enforced separately in run()
 static constexpr int SERVER_TOOL_ISOLATE_EXEC_TIMEOUT = 15; // seconds
 static constexpr size_t SERVER_TOOL_ISOLATE_READ_FILE_MAX_SIZE = 64 * 1024 * 1024; // 64 MB
@@ -524,33 +564,12 @@ public:
    }

    bool write_file(const std::string & path, const std::string & content) const override {
-        std::string abs_path = resolve(path);
-
-        std::error_code ec;
-        fs::path tmp_dir = fs::temp_directory_path(ec);
-        if (ec) return false;
-
-        static std::atomic<uint64_t> tmp_counter{0};
-        fs::path tmp = tmp_dir / string_format(
-            "llama-tools-io-isolate-%zu-%llu.tmp",
-            std::hash<std::thread::id>{}(std::this_thread::get_id()),
-            (unsigned long long) tmp_counter.fetch_add(1));
-
-        {
-            std::ofstream f(tmp, std::ios::binary);
-            if (!f) return false;
-            f << content;
-            if (!f) return false;
-        }
-
-        bool ok = shell_run({"sh", "-c", "mkdir -p \"$(dirname \"$1\")\"", "_", abs_path});
-        if (ok) {
-            ok = upload(tmp.string(), abs_path);
-        }
-
-        std::error_code rm_ec;
-        fs::remove(tmp, rm_ec);
-        return ok;
+        // the content travels on stdin: no argv for the far side to re-parse, no temp file on the host
+        auto res = run_subprocess(
+            build_argv({"sh", "-c", "mkdir -p \"$(dirname \"$1\")\" && cat > \"$1\"", "_", resolve(path)},
+                       /*needs_stdin=*/true),
+            4096, SERVER_TOOL_ISOLATE_EXEC_TIMEOUT, nullptr, true, "", &content);
+        return res.exit_code == 0 && !res.timed_out;
    }

    list_result list_entries(const std::string & base, int max_depth, list_kind kind) const override {
@@ -612,9 +631,6 @@ protected:
    // a transport that re-parses its args in a remote shell (ssh) must join `inner` with shell_quote_join()
    virtual std::vector<std::string> build_argv(const std::vector<std::string> & inner, bool needs_stdin) const = 0;

-    // copy a host file into the isolate, `isolate_path` is absolute and its parent already exists
-    virtual bool upload(const std::string & host_path, const std::string & isolate_path) const = 0;
-
    // quote `argv` into a single string that a POSIX shell re-parses into exactly `argv`
    static std::string shell_quote_join(const std::vector<std::string> & argv) {
        std::string out;
@@ -634,7 +650,7 @@ protected:
 private:
    std::string cwd;

-    // set the working directory in the command itself, docker's `-w` has no equivalent on every transport
+    // set the working directory in the command itself, no `-w` equivalent exists on every transport
    // auxiliary calls do not need this, they use the absolute paths from resolve()
    std::vector<std::string> with_cwd(const std::vector<std::string> & inner) const {
        if (cwd.empty()) {
@@ -697,15 +713,16 @@ private:
    }
 };

-// an already-running docker container, driven through `docker exec` and `docker cp`
-class tools_io_docker : public tools_io_isolate {
+// an already-running container, driven through `<engine> exec`
+// docker and podman take the same verbs and the same argument order, so one class drives both
+class tools_io_container : public tools_io_isolate {
 public:
-    tools_io_docker(std::string container_id, std::string cwd = "")
-        : tools_io_isolate(std::move(cwd)), container_id(std::move(container_id)) {}
+    tools_io_container(std::string bin, std::string container_id, std::string cwd = "")
+        : tools_io_isolate(std::move(cwd)), bin(std::move(bin)), container_id(std::move(container_id)) {}

 protected:
    std::vector<std::string> build_argv(const std::vector<std::string> & inner, bool needs_stdin) const override {
-        std::vector<std::string> argv = {"docker", "exec"};
+        std::vector<std::string> argv = {bin, "exec"};
        if (needs_stdin) {
            argv.push_back("-i");
        }
@@ -714,30 +731,118 @@ protected:
        return argv;
    }

-    bool upload(const std::string & host_path, const std::string & isolate_path) const override {
-        auto res = run_subprocess(
-            {"docker", "cp", host_path, container_id + ":" + isolate_path},
-            4096, SERVER_TOOL_ISOLATE_EXEC_TIMEOUT, nullptr, true);
-        return res.exit_code == 0 && !res.timed_out;
-    }
-
 private:
+    std::string bin;
    std::string container_id;
 };

-// runtime spec used by --tools-runtime and the x-tool-runtime header
-// this is the only scheme for now, ssh: and podman: can be added next to it
-static const std::string SERVER_TOOL_RUNTIME_DOCKER_CONTAINER = "docker-container:";
+// a remote host reached over ssh
+// this is remoting, not isolation: the tools can do anything the target account can do
+class tools_io_ssh : public tools_io_isolate {
+public:
+    tools_io_ssh(std::string target, std::string cwd = "")
+        : tools_io_isolate(std::move(cwd)), target(std::move(target)) {}
+
+    // the target can come from a client header, and ssh reads options from its argv
+    // a target starting with '-' would become one, e.g. -oProxyCommand=<anything> runs on the host
+    static bool is_valid_target(const std::string & target) {
+        if (target.empty() || target[0] == '-') {
+            return false;
+        }
+        return std::all_of(target.begin(), target.end(), [](unsigned char c) {
+            return std::isalnum(c) || c == '.' || c == '-' || c == '_' || c == '@';
+        });
+    }
+
+protected:
+    std::vector<std::string> build_argv(const std::vector<std::string> & inner, bool needs_stdin) const override {
+        // the remote shell re-parses the command line, so `inner` travels as one quoted word
+        std::vector<std::string> argv = ssh_argv();
+        if (!needs_stdin) {
+            argv.push_back("-n");
+        }
+        argv.push_back(target);
+        argv.push_back(shell_quote_join(inner));
+        return argv;
+    }
+
+private:
+    std::string target;
+
+    // there is no console here, so a prompt would hang the tool call
+    // key-based auth only, and the admin must trust the host key beforehand
+    static std::vector<std::string> ssh_argv() {
+        return {
+            "ssh",
+            "-o", "BatchMode=yes",
+            "-o", "PasswordAuthentication=no",
+            "-o", "KbdInteractiveAuthentication=no",
+            "-o", "StrictHostKeyChecking=yes",
+        };
+    }
+};
+
+// "<engine>:<image>" spawns a container and owns it, "<engine>-container:<id>" attaches to one
+struct container_runtime_spec {
+    std::string bin;
+    std::string arg; // image name when spawning, container id when attaching
+    bool attach = false;
+
+    static bool parse(const std::string & spec, container_runtime_spec & out) {
+        // docker and podman take the same verbs, hence a single implementation
+        static const char * engines[] = {"docker", "podman"};
+        for (const char * bin : engines) {
+            const std::string attach_prefix = std::string(bin) + "-container:";
+            if (spec.rfind(attach_prefix, 0) == 0) {
+                out = {bin, spec.substr(attach_prefix.size()), true};
+                return true;
+            }
+            const std::string spawn_prefix = std::string(bin) + ":";
+            if (spec.rfind(spawn_prefix, 0) == 0) {
+                out = {bin, spec.substr(spawn_prefix.size()), false};
+                return true;
+            }
+        }
+        return false;
+    }
+
+    // same risk as the ssh target: an id starting with '-' would become an engine option,
+    // e.g. --privileged
+    static bool is_valid_id(const std::string & id) {
+        if (id.empty() || !std::isalnum((unsigned char) id[0])) {
+            return false;
+        }
+        return std::all_of(id.begin(), id.end(), [](unsigned char c) {
+            return std::isalnum(c) || c == '.' || c == '-' || c == '_';
+        });
+    }
+};

-// an empty runtime runs the tools on the host
 static std::unique_ptr<tools_io> make_tools_io(const json & params) {
    std::string cwd     = json_value(params, "cwd", std::string());
    std::string runtime = json_value(params, "runtime", std::string());
    if (runtime.empty()) {
+        // an empty runtime runs the tools on the host
        return std::make_unique<tools_io_basic>(cwd);
    }
-    if (runtime.rfind(SERVER_TOOL_RUNTIME_DOCKER_CONTAINER, 0) == 0) {
-        return std::make_unique<tools_io_docker>(runtime.substr(SERVER_TOOL_RUNTIME_DOCKER_CONTAINER.size()), cwd);
+    container_runtime_spec container;
+    if (container_runtime_spec::parse(runtime, container)) {
+        // spawning belongs to the runtime that owns the container, a tool call only attaches
+        if (!container.attach) {
+            throw std::runtime_error("tool runtime must name a running container: " + runtime);
+        }
+        if (!container_runtime_spec::is_valid_id(container.arg)) {
+            throw std::runtime_error("invalid container id: " + container.arg);
+        }
+        return std::make_unique<tools_io_container>(container.bin, container.arg, cwd);
+    }
+    const std::string ssh_prefix = "ssh:";
+    if (runtime.rfind(ssh_prefix, 0) == 0) {
+        std::string target = runtime.substr(ssh_prefix.size());
+        if (!tools_io_ssh::is_valid_target(target)) {
+            throw std::runtime_error("invalid ssh target: " + target);
+        }
+        return std::make_unique<tools_io_ssh>(target, cwd);
    }
    // do not fall back to the host, the caller asked for an isolate
    throw std::runtime_error("unknown tool runtime: " + runtime);
@@ -1769,81 +1874,82 @@ struct server_mcp_tool : server_tool {
    }
 };

-// owns the docker container used as the sandboxed runtime for tool invocations, as configured by
-// --tools-runtime. "spawned" mode starts and stops the container itself; "existing" mode just reuses
-// a container id the user already has running and never stops it.
-struct server_tools_docker_runtime {
-    server_tools_docker_runtime(const server_tools_docker_runtime &) = delete;
+// resolves --tools-runtime into the isolate that every tool call runs through
+// spec() returns the runtime string make_tools_io() takes, and runs once per tool call
+struct server_tools_runtime {
+    virtual ~server_tools_runtime() = default;
+    virtual std::string spec() = 0;
+};

-    explicit server_tools_docker_runtime(const std::string & spec) {
-        static const std::string docker_prefix = "docker:";
-        if (spec.rfind(docker_prefix, 0) == 0) {
-            spawned = true;
-            image   = spec.substr(docker_prefix.size());
-            if (image.empty()) {
-                throw std::runtime_error("--tools-runtime docker:<image> requires an image name");
-            }
-            spawn();
-        } else if (spec.rfind(SERVER_TOOL_RUNTIME_DOCKER_CONTAINER, 0) == 0) {
-            spawned      = false;
-            container_id = spec.substr(SERVER_TOOL_RUNTIME_DOCKER_CONTAINER.size());
-            if (container_id.empty()) {
-                throw std::runtime_error("--tools-runtime docker-container:<id> requires a container id");
-            }
-        } else {
+// a target that already exists and needs no lifecycle
+// the spec is validated once at startup, then passed straight through
+struct server_tools_static_runtime : server_tools_runtime {
+    explicit server_tools_static_runtime(std::string spec) : runtime_spec(std::move(spec)) {}
+    std::string spec() override { return runtime_spec; }
+
+private:
+    std::string runtime_spec;
+};
+
+// owns the container the tools run in, as set by --tools-runtime "<engine>:<image>"
+// it is spawned here and stopped when the server exits
+struct server_tools_container_runtime : server_tools_runtime {
+    server_tools_container_runtime(const server_tools_container_runtime &) = delete;
+
+    explicit server_tools_container_runtime(const std::string & spec) {
+        container_runtime_spec parsed;
+        if (!container_runtime_spec::parse(spec, parsed)) {
            throw std::runtime_error("unknown --tools-runtime option: " + spec);
        }
-    }

-    ~server_tools_docker_runtime() {
-        if (spawned && !container_id.empty()) {
-            // closing stdin signals the container's shell (its pid 1) to exit; --rm then removes it
-            proc.close_stdin();
-            proc.join();
+        bin   = parsed.bin;
+        image = parsed.arg;
+        if (image.empty()) {
+            throw std::runtime_error("--tools-runtime " + bin + ":<image> requires an image name");
        }
+        spawn();
    }

-    // container id to use for the next tool call; respawns a spawned container that died on its own,
-    // or throws if an externally-managed one is no longer reachable
-    std::string get_container_id() {
+    ~server_tools_container_runtime() override {
+        // closing stdin signals the container's shell (its pid 1) to exit; --rm then removes it
+        proc.close_stdin();
+        proc.join();
+    }
+
+    // respawns a container that died on its own, so the returned spec always names a running one
+    std::string spec() override {
        std::lock_guard<std::mutex> lock(mutex);
-        if (!spawned) {
-            if (!is_running(container_id)) {
-                throw std::runtime_error(string_format(
-                    "docker container \"%s\" is no longer running, restart it to keep using tools",
-                    container_id.c_str()));
-            }
-            return container_id;
-        }
-
        if (!proc.alive()) {
-            SRV_WRN("docker tools runtime container \"%s\" died, respawning\n", container_id.c_str());
+            SRV_WRN("%s tools runtime container \"%s\" died, respawning\n", bin.c_str(), container_id.c_str());
            spawn();
        }
-        return container_id;
+        return bin + "-container:" + container_id;
    }

 private:
-    bool spawned = false;
-    std::string image; // spawned mode only
+    std::string bin;
+    std::string image;
    std::string container_id;
-    common_subproc proc; // spawned mode only: `docker run` client that keeps the container alive
+    common_subproc proc; // `<engine> run` client that keeps the container alive
    std::mutex mutex;

-    // spawns "docker run --rm -i <image> sh" and keeps its stdin open; the shell blocks reading stdin,
+    // spawns "<engine> run --rm -i <image> sh" and keeps its stdin open; the shell blocks reading stdin,
    // so the container stays alive until we close it (see destructor) or it is killed from the outside
    void spawn() {
+        // create() writes over the handle it is given, so the previous one is released first
+        proc.join();
+
        std::error_code ec;
        fs::path cidfile = fs::temp_directory_path(ec) / string_format(
            "llama-tools-runtime-cid-%zu.tmp", std::hash<std::thread::id>{}(std::this_thread::get_id()));
        fs::remove(cidfile, ec);

-        std::vector<std::string> args = {"docker", "run", "--rm", "-i", "--cidfile", cidfile.string(), image, "sh"};
+        std::vector<std::string> args = {bin, "run", "--rm", "-i", "--cidfile", path_to_utf8(cidfile), image, "sh"};
        int options = subprocess_option_no_window
                    | subprocess_option_inherit_environment
                    | subprocess_option_search_user_path;
        if (!proc.create(args, options)) {
-            throw std::runtime_error("failed to spawn docker container for tools runtime (image: " + image + ")");
+            throw std::runtime_error("failed to spawn " + bin + " container for tools runtime (image: " + image + ")");
        }

        std::string cid;
@@ -1855,15 +1961,10 @@ private:
        fs::remove(cidfile, ec);
        if (cid.empty()) {
            proc.terminate();
-            throw std::runtime_error("timed out waiting for docker container to start (image: " + image + ")");
+            throw std::runtime_error("timed out waiting for " + bin + " container to start (image: " + image + ")");
        }
        container_id = cid;
    }
-
-    static bool is_running(const std::string & id) {
-        auto res = run_subprocess({"docker", "inspect", "-f", "{{.State.Running}}", id}, 16, 5, nullptr, true);
-        return res.exit_code == 0 && !res.timed_out && res.output.rfind("true", 0) == 0;
-    }
 };

 static server_tool & find_tool(std::vector<std::unique_ptr<server_tool>> & tools, const std::string & name, bool require_stream) {
@@ -1914,11 +2015,22 @@ static std::string get_header(const std::map<std::string, std::string> & headers
 server_tools::server_tools() = default;
 server_tools::~server_tools() = default;

+// the "<engine>:<image>" form owns a container lifecycle
+// anything else names an existing target, so only its spec is validated here at startup
+static std::unique_ptr<server_tools_runtime> make_tools_runtime(const std::string & spec) {
+    container_runtime_spec parsed;
+    if (container_runtime_spec::parse(spec, parsed) && !parsed.attach) {
+        return std::make_unique<server_tools_container_runtime>(spec);
+    }
+    make_tools_io({{"runtime", spec}}); // nothing to own, just reject a bad spec now
+    return std::make_unique<server_tools_static_runtime>(spec);
+}
+
 void server_tools::setup(const std::vector<std::string> & enabled_tools,
                         server_mcp & mcp_mgr,
                         const std::string & tools_runtime) {
    if (!tools_runtime.empty()) {
-        docker_runtime = std::make_unique<server_tools_docker_runtime>(tools_runtime);
+        runtime = make_tools_runtime(tools_runtime);
    }

    if (!enabled_tools.empty()) {
@@ -2016,11 +2128,11 @@ void server_tools::setup(const std::vector<std::string> & enabled_tools,
            if (params.contains("runtime")) {
                params.erase("runtime");
            }
-            auto runtime = get_header(req.headers, "x-tool-runtime");
-            if (!runtime.empty()) {
-                params["runtime"] = runtime;
-            } else if (docker_runtime) {
-                params["runtime"] = SERVER_TOOL_RUNTIME_DOCKER_CONTAINER + docker_runtime->get_container_id();
+            auto runtime_header = get_header(req.headers, "x-tool-runtime");
+            if (!runtime_header.empty()) {
+                params["runtime"] = runtime_header;
+            } else if (runtime) {
+                params["runtime"] = runtime->spec();
            }

            server_tool & tool = find_tool(tools, tool_name, stream);
@@ -31,7 +31,7 @@ struct server_tool {
    json to_json() const;
 };

-struct server_tools_docker_runtime; // impl detail, defined in server-tools.cpp
+struct server_tools_runtime; // impl detail, defined in server-tools.cpp

 struct server_tools {
    std::vector<std::unique_ptr<server_tool>> tools;
@@ -40,8 +40,8 @@ struct server_tools {
    server_response queue_res;
    std::atomic<int> res_id{0};

-    // set when --tools-runtime is configured; owns the docker container used to run tools, if any
-    std::unique_ptr<server_tools_docker_runtime> docker_runtime;
+    // set when --tools-runtime is configured; routes every tool call through an isolate
+    std::unique_ptr<server_tools_runtime> runtime;

    void setup(const std::vector<std::string> & enabled_tools,
               server_mcp & mcp_mgr,
@@ -89,7 +89,7 @@ int llama_server(int argc, char ** argv) {
    std::setlocale(LC_NUMERIC, "C");

 #ifndef _WIN32
-    // Ignore SIGPIPE so the server does not crash if an MCP child exits while we are writing to its stdin
+    // Ignore SIGPIPE so the server does not crash if a child (MCP server, tools runtime) exits while we are writing to its stdin
    signal(SIGPIPE, SIG_IGN);
 #endif

@@ -14,7 +14,7 @@ PROJECT_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..
 GREP_MARKER = "llama_cpp_test_tools_builtin_marker_grep_search"

 # image the container runtime tests run their shell in
-DOCKER_IMAGE = "busybox"
+CONTAINER_IMAGE = "busybox"


@pytest.fixture(autouse=True)
@@ -151,54 +151,59 @@ def test_tools_builtin_cwd_header():
            os.remove(marker_path)


-def _docker_unavailable_reason() -> str | None:
-    """None if docker can run the image these tests use, otherwise the reason it can't."""
-    docker_bin = shutil.which("docker")
-    if docker_bin is None:
-        return "docker is not installed"
+def _container_engine_unavailable_reason(engine: str) -> str | None:
+    """None if `engine` can run the image these tests use, otherwise the reason it can't."""
+    engine_bin = shutil.which(engine)
+    if engine_bin is None:
+        return f"{engine} is not installed"
    try:
-        # a daemon that answers `docker info` still cannot run a linux image when it serves
-        # windows containers, so probe the image itself, which also pulls it before the tests
-        subprocess.run([docker_bin, "run", "--rm", DOCKER_IMAGE, "true"], capture_output=True, timeout=60, check=True)
+        # a daemon that answers `info` still cannot run a linux image when it serves windows
+        # containers, so probe the image itself, which also pulls it before the tests
+        subprocess.run([engine_bin, "run", "--rm", CONTAINER_IMAGE, "true"], capture_output=True, timeout=60, check=True)
    except Exception as e:
-        return f"docker cannot run {DOCKER_IMAGE}: {e}"
+        return f"{engine} cannot run {CONTAINER_IMAGE}: {e}"
    return None


-@pytest.fixture
-def docker_container():
-    reason = _docker_unavailable_reason()
+@pytest.fixture(params=["docker", "podman"])
+def container_engine(request):
+    engine = request.param
+    reason = _container_engine_unavailable_reason(engine)
    if reason is not None:
        pytest.skip(reason)  # ty: ignore[too-many-positional-arguments, invalid-argument-type]
+    return engine

+
+@pytest.fixture
+def container_id(container_engine: str):
    proc = subprocess.run(
-        ["docker", "run", "-d", "--rm", DOCKER_IMAGE, "sleep", "300"],
+        [container_engine, "run", "-d", "--rm", CONTAINER_IMAGE, "sleep", "300"],
        capture_output=True, text=True,
    )
    if proc.returncode != 0:
-        pytest.skip(f"failed to start docker container: {proc.stderr.strip()}")  # ty: ignore[too-many-positional-arguments, invalid-argument-type]
+        pytest.skip(f"failed to start {container_engine} container: {proc.stderr.strip()}")  # ty: ignore[too-many-positional-arguments, invalid-argument-type]

-    container_id = proc.stdout.strip()
+    cid = proc.stdout.strip()
    try:
-        yield container_id
+        yield cid
    finally:
-        subprocess.run(["docker", "rm", "-f", container_id], capture_output=True)
+        subprocess.run([container_engine, "rm", "-f", cid], capture_output=True)


-def test_tools_builtin_runtime_header(docker_container: str):
+def test_tools_builtin_runtime_header(container_engine: str, container_id: str):
    global server
    server.start()

-    headers = {"x-tool-runtime": f"docker-container:{docker_container}", "x-tool-cwd": "/tmp"}
+    headers = {"x-tool-runtime": f"{container_engine}-container:{container_id}", "x-tool-cwd": "/tmp"}

-    write_res = call_tool("write_file", {"path": "test.log", "content": "hello docker\n"}, headers=headers)
+    write_res = call_tool("write_file", {"path": "test.log", "content": "hello container\n"}, headers=headers)
    assert write_res["result"] == "file written successfully"

    read_res = call_tool("read_file", {"path": "test.log"}, headers=headers)
-    assert read_res["plain_text_response"] == "hello docker\n"
+    assert read_res["plain_text_response"] == "hello container\n"

    exec_res = call_tool("exec_shell_command", {"command": "cat test.log"}, headers=headers)
-    assert "hello docker" in exec_res["plain_text_response"]
+    assert "hello container" in exec_res["plain_text_response"]


 def test_tools_builtin_runtime_header_unknown_scheme():
@@ -208,18 +213,46 @@ def test_tools_builtin_runtime_header_unknown_scheme():
    # an unknown runtime must fail, never silently fall back to running on the host
    res = server.make_request("POST", "/tools",
                              data={"tool": "exec_shell_command", "params": {"command": "echo hi"}},
-                              headers={"x-tool-runtime": "ssh:example.com"})
+                              headers={"x-tool-runtime": "fake:does-not-exist"})
    assert res.status_code == 500, res.body
    assert "unknown tool runtime" in str(res.body)


+def test_tools_builtin_runtime_header_rejects_ssh_option_injection():
+    global server
+    server.start()
+
+    # ssh reads options from its argv, so a target starting with '-' must be rejected
+    res = server.make_request("POST", "/tools",
+                              data={"tool": "exec_shell_command", "params": {"command": "echo hi"}},
+                              headers={"x-tool-runtime": "ssh:-oProxyCommand=touch /tmp/pwned"})
+    assert res.status_code == 500, res.body
+    assert "invalid ssh target" in str(res.body)
+
+
+@pytest.mark.parametrize("engine", ["docker", "podman"])
+def test_tools_builtin_runtime_header_rejects_container_option_injection(engine: str):
+    global server
+    server.start()
+
+    # the container id lands on the `<engine> exec` command line, so an id that looks
+    # like an option must be rejected
+    res = server.make_request("POST", "/tools",
+                              data={"tool": "exec_shell_command", "params": {"command": "echo hi"}},
+                              headers={"x-tool-runtime": f"{engine}-container:--privileged"})
+    assert res.status_code == 500, res.body
+    assert "invalid container id" in str(res.body)
+
+
 def test_tools_builtin_docker_runtime_cleans_up_spawned_container():
-    reason = _docker_unavailable_reason()
+    # docker-only: this reads the container hostname to get the spawned id, which only docker
+    # sets to the short id. podman is covered by the attach path above
+    reason = _container_engine_unavailable_reason("docker")
    if reason is not None:
        pytest.skip(reason)  # ty: ignore[too-many-positional-arguments, invalid-argument-type]

    global server
-    server.server_tools_runtime = f"docker:{DOCKER_IMAGE}"
+    server.server_tools_runtime = f"docker:{CONTAINER_IMAGE}"
    server.start()

    # exec_shell_command runs inside the container spawned for --tools-runtime; docker sets
@@ -797,7 +797,7 @@
 		data-placeholder={placeholder}
 		tabindex={disabled ? -1 : 0}
 		class={[
-			'chat-form-contenteditable text-md min-h-12 w-full whitespace-pre-wrap wrap-break-word border-0 bg-transparent p-0 leading-6 outline-none focus-visible:ring-0 focus-visible:ring-offset-0',
+			'chat-form-contenteditable text-md min-h-12 w-full overflow-y-auto whitespace-pre-wrap wrap-break-word border-0 bg-transparent p-0 leading-6 outline-none focus-visible:ring-0 focus-visible:ring-offset-0',
 			disabled && 'cursor-not-allowed'
 		]}
 		style="max-height: var(--max-message-height);"
@@ -164,7 +164,7 @@
 								? `max-height: ${MAX_HEIGHT}px;`
 								: 'max-height: none;'}
 						>
-							{#if !currentConfig.renderContentAsRawText}
+							{#if currentConfig.renderUserContentAsMarkdown}
 								<div bind:this={messageElement} class={isExpanded ? 'cursor-text' : ''}>
 									<MarkdownContent class="markdown-system-content" content={message.content} />
 								</div>
@@ -1,5 +1,5 @@
 <script lang="ts">
-	import { ChatAttachmentsList, MarkdownContent } from '$lib/components/app';
+	import { ChatAttachmentsList, MarkdownContent, MentionText } from '$lib/components/app';
 	import { Card } from '$lib/components/ui/card';
 	import { config } from '$lib/stores/settings.svelte';
 	import type { DatabaseMessageExtra } from '$lib/types/database';
@@ -64,14 +64,14 @@
 		data-multiline={isMultiline ? '' : undefined}
 		style="{maxHeightStyle} overflow-wrap: anywhere; word-break: break-word;"
 	>
-		{#if renderMarkdown && !currentConfig.renderContentAsRawText}
+		{#if renderMarkdown && currentConfig.renderUserContentAsMarkdown}
 			<div bind:this={messageElement}>
 				<MarkdownContent class="markdown-user-content" {content} />
 			</div>
 		{:else}
-			<span bind:this={messageElement} class="text-md whitespace-pre-wrap">
-				{content}
-			</span>
+			<span bind:this={messageElement} class="text-md whitespace-pre-wrap"
+				><MentionText {content} /></span
+			>
 		{/if}
 	</Card>
 {/if}
@@ -140,7 +140,7 @@
 		class:is-streaming={isPending}
 		onscroll={handleScrollEvent}
 	>
-		{#if !currentConfig.renderContentAsRawText}
+		{#if currentConfig.renderThinkingAsMarkdown}
 			<MarkdownContent content={section.content} class="text-muted-foreground" {attachments} />
 		{:else}
 			<div
@@ -0,0 +1,36 @@
+<script lang="ts">
+	import { SETTINGS_KEYS } from '$lib/constants';
+	import { settingsStore } from '$lib/stores/settings.svelte';
+	import { toolsStore } from '$lib/stores/tools.svelte';
+	import {
+		getMentionBadgeIconPaths,
+		getMentionBadgeLabel,
+		MENTION_BADGE_CLASSNAME,
+		MENTION_BADGE_ICON_CLASSNAME,
+		MENTION_BADGE_SVG_ATTRIBUTES
+	} from '$lib/utils';
+
+	interface Props {
+		name: string;
+		path: string;
+	}
+
+	let { name, path }: Props = $props();
+
+	let showFullPath = $derived(
+		settingsStore.getConfig(SETTINGS_KEYS.SHOW_FULL_PATH_IN_MENTIONS) as boolean
+	);
+	let label = $derived(getMentionBadgeLabel(name, path, showFullPath, toolsStore.serverHome));
+</script>
+
+<!-- The chip is a flex container, so template whitespace between its
+     children collapses away and the icon keeps its `gap-1` spacing. -->
+<span class={MENTION_BADGE_CLASSNAME} title={path}>
+	<svg {...MENTION_BADGE_SVG_ATTRIBUTES} class={MENTION_BADGE_ICON_CLASSNAME}>
+		{#each getMentionBadgeIconPaths(path) as d (d)}
+			<path {d} />
+		{/each}
+	</svg>
+
+	<span class="shrink-0 truncate">{label}</span>
+</span>
@@ -0,0 +1,17 @@
+<script lang="ts">
+	import MentionBadge from './MentionBadge.svelte';
+	import { splitMentionSegments } from '$lib/utils';
+
+	interface Props {
+		content: string;
+	}
+
+	let { content }: Props = $props();
+
+	let segments = $derived(splitMentionSegments(content));
+</script>
+
+<!-- Segments sit in a `whitespace-pre-wrap` parent, so the markup stays
+     glued: any newline between the tags below would print as a space. -->
+<!-- prettier-ignore -->
+{#each segments as segment, index (index)}{#if segment.mention}<MentionBadge name={segment.mention.name} path={segment.mention.path} />{:else}{segment.text}{/if}{/each}
@@ -31,6 +31,20 @@
 */
 export { default as MarkdownContent } from './MarkdownContent/MarkdownContent.svelte';

+/**
+ * **MentionText** - Plain text with file mention badges
+ *
+ * Renders a message verbatim, turning only `[name](file://path)` links
+ * into the same badge chips the markdown path draws. Nothing else is
+ * interpreted, so pasted code keeps its `#` comments and underscores.
+ *
+ * @example
+ * ```svelte
+ * <span class="whitespace-pre-wrap"><MentionText content={message.content} /></span>
+ * ```
+ */
+export { default as MentionText } from './MentionText.svelte';
+
 /**
 * **SyntaxHighlightedCode** - Code syntax highlighting
 *
@@ -10,6 +10,9 @@ export const MENTION_BADGE_CLASSNAME =

 export const MENTION_BADGE_ICON_CLASSNAME = 'h-3 w-3 shrink-0';

+/** Regex flag that makes the mention scanner walk every link in a message instead of the first. */
+export const MENTION_LINK_SCAN_FLAGS = 'g';
+
 /**
 * SVG attributes shared by the DOM-built and hast-built badge icons.
 * The tokenizer applies them via `setAttribute`, the rehype plugin
@@ -12,6 +12,9 @@ import { UrlProtocol } from '$lib/enums';
 export const CWD_CHANGED_PREFIX = 'Set working directory to ';
 export const CWD_CLEARED_TEXT = 'Working directory cleared';

+/** Trailing separator that marks a path as a directory. */
+export const DIRECTORY_PATH_SUFFIX = '/';
+
 export const HOME_TILDE = '~';
 export const HOME_TILDE_PREFIX = '~/'; // tilde plus path separator

@@ -41,7 +41,8 @@ export const SETTINGS_KEYS = {
 	// Performance
 	PRE_ENCODE_CONVERSATION: 'preEncodeConversation',
 	PRESENCE_PENALTY: 'presence_penalty',
-	RENDER_CONTENT_AS_RAW_TEXT: 'renderContentAsRawText',
+	RENDER_THINKING_AS_MARKDOWN: 'renderThinkingAsMarkdown',
+	RENDER_USER_CONTENT_AS_MARKDOWN: 'renderUserContentAsMarkdown',
 	// Penalties
 	REPEAT_LAST_N: 'repeat_last_n',
 	REPEAT_PENALTY: 'repeat_penalty',
@@ -230,10 +230,18 @@ const SETTINGS_REGISTRY: Record<string, SettingsSectionEntry> = {
 				type: SettingsFieldType.CHECKBOX
 			},
 			{
-				defaultValue: false,
-				help: 'Display user, system and thinking content as plain text instead of formatted Markdown. Markdown is the default so that @-mention badges render in sent messages.',
-				key: SETTINGS_KEYS.RENDER_CONTENT_AS_RAW_TEXT,
-				label: 'Render content as raw text',
+				defaultValue: true,
+				help: 'Render user messages using markdown formatting in the chat. Turn this off to keep a message exactly as typed; @-mention badges show either way.',
+				key: SETTINGS_KEYS.RENDER_USER_CONTENT_AS_MARKDOWN,
+				label: 'Render user content as Markdown',
+				section: SETTINGS_SECTION_SLUGS.DISPLAY,
+				type: SettingsFieldType.CHECKBOX
+			},
+			{
+				defaultValue: true,
+				help: 'Render the reasoning/thinking block content as formatted Markdown instead of plain text.',
+				key: SETTINGS_KEYS.RENDER_THINKING_AS_MARKDOWN,
+				label: 'Render thinking as Markdown',
 				section: SETTINGS_SECTION_SLUGS.DISPLAY,
 				type: SettingsFieldType.CHECKBOX
 			},
@@ -629,6 +629,37 @@ const configTypesMigration: Migration = {
 			console.log(`[Migration] Config types: coerced string booleans (changed=${changed})`);
 	}
 };
+const RENDER_KEYS_MIGRATION_ID = 'render-keys-unfold-v1';
+const LEGACY_RENDER_RAW_TEXT_KEY = 'renderContentAsRawText';
+const renderKeysMigration: Migration = {
+	description: 'Unfold the single raw text render toggle onto the per-surface render keys',
+	id: RENDER_KEYS_MIGRATION_ID,
+
+	async run(): Promise<void> {
+		const configRaw = localStorage.getItem(CONFIG_LOCALSTORAGE_KEY);
+
+		if (configRaw === null) return;
+
+		const config = JSON.parse(configRaw);
+
+		if (!(LEGACY_RENDER_RAW_TEXT_KEY in config)) return;
+
+		// The toggle carried user content and thinking at once and cannot say which surface
+		// was chosen, so it only restores the user key and thinking keeps its own default.
+		if (!(SETTINGS_KEYS.RENDER_USER_CONTENT_AS_MARKDOWN in config)) {
+			config[SETTINGS_KEYS.RENDER_USER_CONTENT_AS_MARKDOWN] =
+				config[LEGACY_RENDER_RAW_TEXT_KEY] !== true;
+		}
+
+		// Dropped rather than preserved: the two render keys and the toggle describe the same
+		// surfaces, so leaving it behind would let a stale value fight the restored one.
+		delete config[LEGACY_RENDER_RAW_TEXT_KEY];
+		localStorage.setItem(CONFIG_LOCALSTORAGE_KEY, JSON.stringify(config));
+
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log('[Migration] Render keys: unfolded the raw text toggle');
+	}
+};
 const MCP_DEFAULT_OVERRIDES_LEGACY_KEY = `${STORAGE_APP_NAME}.mcpDefaultServerOverrides`;
 const MCP_DEFAULT_OVERRIDES_MERGE_MIGRATION_ID = 'mcp-default-overrides-merge-v1';
 /**
@@ -722,7 +753,8 @@ const migrations: Migration[] = [
 	customJsonKeyMigration,
 	mcpDefaultEnabledMigration,
 	mcpDefaultOverridesMergeMigration,
-	configTypesMigration
+	configTypesMigration,
+	renderKeysMigration
 ];

 export const MigrationService = {
@@ -135,32 +135,6 @@ class SettingsStore {
 				...savedVal
 			};

-			// Migrate the legacy render keys into `renderContentAsRawText`
-			// (inverted semantics: the old keys opted INTO markdown). Any
-			// explicit raw-text preference wins when the legacy keys disagree.
-			const LEGACY_MARKDOWN_KEYS = ['renderUserContentAsMarkdown', 'renderThinkingAsMarkdown'];
-			const LEGACY_RAW_TEXT_KEY = 'renderUserContentAsRawText'; // this branch's intermediate key
-			const legacyKeys = [...LEGACY_MARKDOWN_KEYS, LEGACY_RAW_TEXT_KEY].filter(
-				(key) => key in savedVal
-			);
-
-			if (legacyKeys.length > 0) {
-				if (!(SETTINGS_KEYS.RENDER_CONTENT_AS_RAW_TEXT in savedVal)) {
-					if (LEGACY_RAW_TEXT_KEY in savedVal) {
-						this.config[SETTINGS_KEYS.RENDER_CONTENT_AS_RAW_TEXT] = savedVal[LEGACY_RAW_TEXT_KEY];
-					} else {
-						this.config[SETTINGS_KEYS.RENDER_CONTENT_AS_RAW_TEXT] = LEGACY_MARKDOWN_KEYS.filter(
-							(key) => key in savedVal
-						).some((key) => savedVal[key] === false);
-					}
-				}
-
-				for (const key of legacyKeys) {
-					delete (this.config as Record<string, unknown>)[key];
-				}
-				this.saveConfig();
-			}
-
 			// Default sendOnEnter to false on mobile when the user has no saved preference
 			if (!(SETTINGS_KEYS.SEND_ON_ENTER in savedVal)) {
 				if (isMobile.current) {
@@ -240,6 +240,7 @@ export {
 	MENTION_BADGE_FOLDER_ICON_PATHS,
 	getMentionBadgeIconPaths,
 	getMentionBadgeLabel,
+	splitMentionSegments,
 	buildMentionInsertion
 } from './mention-badge';

@@ -1,8 +1,9 @@
 import { abbreviateHome, lastPathSegment } from './path-display';
-import { FILE_URI_PREFIX } from '$lib/constants';
+import { DIRECTORY_PATH_SUFFIX, FILE_URI_PREFIX } from '$lib/constants';
 import {
 	MENTION_BADGE_FILE_ICON_PATHS,
-	MENTION_BADGE_FOLDER_ICON_PATHS
+	MENTION_BADGE_FOLDER_ICON_PATHS,
+	MENTION_LINK_SCAN_FLAGS
 } from '$lib/constants/mention-badge';
 import { FileMentionEntryType } from '$lib/enums';
 import type { FileMentionEntry } from '$lib/types';
@@ -48,8 +49,45 @@ export function decodeFileLinkPath(path: string): string {
 	}
 }

+export interface MentionTextSegment {
+	text: string;
+	mention: { name: string; path: string } | null;
+}
+
+/**
+ * Split raw text into plain runs and `[name](file://path)` mentions.
+ * The raw-text renderers walk these segments to draw badges without
+ * handing the message to the markdown parser, so a `#` stays a `#`.
+ */
+export function splitMentionSegments(value: string): MentionTextSegment[] {
+	const linkRe = fileMentionLinkRe(MENTION_LINK_SCAN_FLAGS);
+	const segments: MentionTextSegment[] = [];
+
+	let cursor = 0;
+	let match: RegExpExecArray | null;
+
+	while ((match = linkRe.exec(value)) !== null) {
+		if (match.index > cursor) {
+			segments.push({ mention: null, text: value.slice(cursor, match.index) });
+		}
+
+		segments.push({
+			mention: { name: match[1], path: decodeFileLinkPath(match[2]) },
+			text: match[0]
+		});
+
+		cursor = match.index + match[0].length;
+	}
+
+	if (cursor < value.length) segments.push({ mention: null, text: value.slice(cursor) });
+
+	return segments;
+}
+
 export function getMentionBadgeIconPaths(path: string): readonly string[] {
-	return path.endsWith('/') ? MENTION_BADGE_FOLDER_ICON_PATHS : MENTION_BADGE_FILE_ICON_PATHS;
+	return path.endsWith(DIRECTORY_PATH_SUFFIX)
+		? MENTION_BADGE_FOLDER_ICON_PATHS
+		: MENTION_BADGE_FILE_ICON_PATHS;
 }

 export function getMentionBadgeLabel(
@@ -1,66 +0,0 @@
-// Guards the legacy render-key migration: `renderUserContentAsMarkdown`
-// and `renderThinkingAsMarkdown` (opt-INTO markdown) fold into the single
-// `renderContentAsRawText` setting, with any explicit raw-text preference
-// winning when the legacy keys disagree. Legacy keys are removed from the
-// persisted config so they do not stay orphaned in localStorage.
-
-import { CONFIG_LOCALSTORAGE_KEY } from '$lib/constants/storage';
-import { config, settingsStore } from '$lib/stores/settings.svelte';
-import { beforeEach, describe, expect, it } from 'vitest';
-
-function seedConfig(stored: Record<string, unknown>) {
-	localStorage.setItem(CONFIG_LOCALSTORAGE_KEY, JSON.stringify(stored));
-	settingsStore.initialize();
-}
-
-function persisted(): Record<string, unknown> {
-	return JSON.parse(localStorage.getItem(CONFIG_LOCALSTORAGE_KEY) ?? '{}');
-}
-
-describe('renderContentAsRawText migration', () => {
-	beforeEach(() => {
-		localStorage.removeItem(CONFIG_LOCALSTORAGE_KEY);
-		settingsStore.initialize();
-	});
-
-	it('maps renderUserContentAsMarkdown=false to raw text', () => {
-		seedConfig({ renderUserContentAsMarkdown: false });
-		expect(config().renderContentAsRawText).toBe(true);
-	});
-
-	it('maps renderUserContentAsMarkdown=true to markdown', () => {
-		seedConfig({ renderUserContentAsMarkdown: true });
-		expect(config().renderContentAsRawText).toBe(false);
-	});
-
-	it('maps renderThinkingAsMarkdown=false to raw text', () => {
-		seedConfig({ renderThinkingAsMarkdown: false });
-		expect(config().renderContentAsRawText).toBe(true);
-	});
-
-	it('lets any explicit raw-text preference win when the legacy keys disagree', () => {
-		seedConfig({ renderThinkingAsMarkdown: false, renderUserContentAsMarkdown: true });
-		expect(config().renderContentAsRawText).toBe(true);
-	});
-
-	it('honors the intermediate renderUserContentAsRawText key from the PR branch', () => {
-		seedConfig({ renderUserContentAsRawText: true });
-		expect(config().renderContentAsRawText).toBe(true);
-	});
-
-	it('keeps an already-migrated value and cleans up the legacy keys', () => {
-		seedConfig({ renderContentAsRawText: false, renderUserContentAsMarkdown: false });
-		expect(config().renderContentAsRawText).toBe(false);
-
-		const stored = persisted();
-
-		expect(stored.renderUserContentAsMarkdown).toBeUndefined();
-		expect(stored.renderThinkingAsMarkdown).toBeUndefined();
-		expect(stored.renderUserContentAsRawText).toBeUndefined();
-	});
-
-	it('defaults to markdown when no legacy key exists', () => {
-		seedConfig({});
-		expect(config().renderContentAsRawText).toBe(false);
-	});
-});
@@ -0,0 +1,64 @@
+// Guards the unfolding of `renderContentAsRawText` back onto the two
+// per-surface render keys. The single toggle carried user content and
+// thinking at once, so only the user key is restored from it and thinking
+// returns to its own default. The toggle is removed from the persisted
+// config so it does not stay orphaned in localStorage.
+
+import { CONFIG_LOCALSTORAGE_KEY } from '$lib/constants/storage';
+import { MigrationService } from '$lib/services/migration.service';
+import { config, settingsStore } from '$lib/stores/settings.svelte';
+import { beforeEach, describe, expect, it } from 'vitest';
+
+const RENDER_KEYS_MIGRATION_ID = 'render-keys-unfold-v1';
+
+async function seedConfig(stored: Record<string, unknown>) {
+	localStorage.setItem(CONFIG_LOCALSTORAGE_KEY, JSON.stringify(stored));
+
+	const migration = MigrationService.getMigrations().find((m) => m.id === RENDER_KEYS_MIGRATION_ID);
+
+	await migration?.run();
+	settingsStore.initialize();
+}
+
+function persisted(): Record<string, unknown> {
+	return JSON.parse(localStorage.getItem(CONFIG_LOCALSTORAGE_KEY) ?? '{}');
+}
+
+describe('renderContentAsRawText unfolding', () => {
+	beforeEach(() => {
+		localStorage.removeItem(CONFIG_LOCALSTORAGE_KEY);
+		MigrationService.resetState();
+		settingsStore.initialize();
+	});
+
+	it('maps raw text to user content as plain text', async () => {
+		await seedConfig({ renderContentAsRawText: true });
+		expect(config().renderUserContentAsMarkdown).toBe(false);
+	});
+
+	it('maps markdown to user content as markdown', async () => {
+		await seedConfig({ renderContentAsRawText: false });
+		expect(config().renderUserContentAsMarkdown).toBe(true);
+	});
+
+	it('leaves thinking on its own default', async () => {
+		await seedConfig({ renderContentAsRawText: true });
+		expect(config().renderThinkingAsMarkdown).toBe(true);
+	});
+
+	it('keeps an explicit user preference over the toggle', async () => {
+		await seedConfig({ renderContentAsRawText: true, renderUserContentAsMarkdown: true });
+		expect(config().renderUserContentAsMarkdown).toBe(true);
+	});
+
+	it('drops the toggle from the persisted config', async () => {
+		await seedConfig({ renderContentAsRawText: true });
+		expect(persisted().renderContentAsRawText).toBeUndefined();
+	});
+
+	it('leaves both surfaces on markdown when nothing is stored', async () => {
+		await seedConfig({});
+		expect(config().renderUserContentAsMarkdown).toBe(true);
+		expect(config().renderThinkingAsMarkdown).toBe(true);
+	});
+});
@@ -0,0 +1,48 @@
+import { splitMentionSegments } from '$lib/utils/mention-badge';
+import { describe, expect, it } from 'vitest';
+
+describe('splitMentionSegments', () => {
+	it('returns a single plain run when there is no mention', () => {
+		const segments = splitMentionSegments('# not a heading here');
+
+		expect(segments).toEqual([{ mention: null, text: '# not a heading here' }]);
+	});
+
+	it('splits text around a mention', () => {
+		const segments = splitMentionSegments('look at [main.c](file:///src/main.c) please');
+
+		expect(segments.map((segment) => segment.text)).toEqual([
+			'look at ',
+			'[main.c](file:///src/main.c)',
+			' please'
+		]);
+		expect(segments[1].mention).toEqual({ name: 'main.c', path: '/src/main.c' });
+	});
+
+	it('decodes percent-encoded paths', () => {
+		const segments = splitMentionSegments('[a b.txt](file:///tmp/a%20b.txt)');
+
+		expect(segments[0].mention?.path).toBe('/tmp/a b.txt');
+	});
+
+	it('keeps the directory marker so the folder icon is picked', () => {
+		const segments = splitMentionSegments('[src](file:///repo/src/)');
+
+		expect(segments[0].mention?.path).toBe('/repo/src/');
+	});
+
+	it('handles adjacent mentions with no text between them', () => {
+		const segments = splitMentionSegments('[a](file:///a)[b](file:///b)');
+
+		expect(segments).toHaveLength(2);
+		expect(segments.every((segment) => segment.mention !== null)).toBe(true);
+	});
+
+	it('preserves the exact source when segments are joined back', () => {
+		const source = 'see [a](file:///a) and [b](file:///b/) done';
+
+		expect(splitMentionSegments(source).reduce((acc, segment) => acc + segment.text, '')).toBe(
+			source
+		);
+	});
+});
@@ -8,8 +8,8 @@
 #ifndef CPPHTTPLIB_HTTPLIB_H
 #define CPPHTTPLIB_HTTPLIB_H

-#define CPPHTTPLIB_VERSION "0.52.0"
-#define CPPHTTPLIB_VERSION_NUM "0x003400"
+#define CPPHTTPLIB_VERSION "0.53.0"
+#define CPPHTTPLIB_VERSION_NUM "0x003500"

 #ifdef _WIN32
 #if defined(_WIN32_WINNT) && _WIN32_WINNT < 0x0A00
@@ -1805,6 +1805,7 @@ enum class Error {
  HTTPParsing,
  InvalidRangeHeader,
  UnsupportedContentEncoding,
+  WebSocketHandshake,

  // For internal use only
  SSLPeerCouldBeClosed_,
@@ -3233,8 +3234,6 @@ private:
  // Used to keep custom CA configuration exclusive with system CA loading.
  bool ca_cert_store_set_ = false;

-  long verify_result_ = 0;
-
  std::function<SSLVerifierResponse(tls::session_t)> session_verifier_;

 #ifdef CPPHTTPLIB_WINDOWS_AUTOMATIC_ROOT_CERTIFICATES_UPDATE
@@ -4204,6 +4203,50 @@ enum class CloseStatus : uint16_t {

 enum ReadResult : int { Fail = 0, Text = 1, Binary = 2 };

+// Result of WebSocketClient::connect(). Truthy only when the WebSocket
+// upgrade handshake fully succeeded. On failure error() identifies the
+// failing layer; status()/headers() expose the server's upgrade response
+// when one was received (status() is -1 otherwise).
+class Result {
+public:
+  Result() = default;
+  Result(Error err, int status, Headers &&headers)
+      : err_(err), status_(status), headers_(std::move(headers)) {}
+
+  explicit operator bool() const { return err_ == Error::Success; }
+  Error error() const { return err_; }
+
+  // Upgrade response info
+  int status() const { return status_; }
+  const Headers &headers() const { return headers_; }
+  std::string get_header_value(const std::string &key,
+                               const char *def = "") const {
+    return detail::get_header_value(headers_, key, def, 0);
+  }
+  bool has_header(const std::string &key) const {
+    return headers_.find(key) != headers_.end();
+  }
+
+#ifdef CPPHTTPLIB_SSL_ENABLED
+  Result(Error err, int status, Headers &&headers, int ssl_error,
+         uint64_t ssl_backend_error)
+      : err_(err), status_(status), headers_(std::move(headers)),
+        ssl_error_(ssl_error), ssl_backend_error_(ssl_backend_error) {}
+
+  int ssl_error() const { return ssl_error_; }
+  uint64_t ssl_backend_error() const { return ssl_backend_error_; }
+#endif
+
+private:
+  Error err_ = Error::Unknown; // a default-constructed Result is falsy
+  int status_ = -1;
+  Headers headers_;
+#ifdef CPPHTTPLIB_SSL_ENABLED
+  int ssl_error_ = 0;
+  uint64_t ssl_backend_error_ = 0;
+#endif
+};
+
 class WebSocket {
 public:
  WebSocket(const WebSocket &) = delete;
@@ -4270,7 +4313,7 @@ public:

  bool is_valid() const;

-  bool connect();
+  Result connect();
  ReadResult read(std::string &msg);
  bool send(const std::string &data);
  bool send(const char *data, size_t len);
@@ -4279,28 +4322,52 @@ public:
  bool is_open() const;
  const std::string &subprotocol() const;
  void set_read_timeout(time_t sec, time_t usec = 0);
+  template <class Rep, class Period>
+  void set_read_timeout(const std::chrono::duration<Rep, Period> &duration);
+
  void set_write_timeout(time_t sec, time_t usec = 0);
+  template <class Rep, class Period>
+  void set_write_timeout(const std::chrono::duration<Rep, Period> &duration);
+
  void set_websocket_ping_interval(time_t sec);
  void set_websocket_max_missed_pongs(int count);
  void set_tcp_nodelay(bool on);
  void set_address_family(int family);
  void set_ipv6_v6only(bool on);
  void set_socket_options(SocketOptions socket_options);
+
  void set_connection_timeout(time_t sec, time_t usec = 0);
+  template <class Rep, class Period>
+  void
+  set_connection_timeout(const std::chrono::duration<Rep, Period> &duration);
+
  void set_interface(const std::string &intf);
  void set_hostname_addr_map(std::map<std::string, std::string> addr_map);

 #ifdef CPPHTTPLIB_SSL_ENABLED
-  void set_ca_cert_path(const std::string &path);
+  struct PemMemory {
+    const char *cert_pem;
+    size_t cert_pem_len;
+    const char *key_pem;
+    size_t key_pem_len;
+    const char *private_key_password;
+  };
+  explicit WebSocketClient(const std::string &scheme_host_port_path,
+                           const PemMemory &pem, const Headers &headers = {});
+
+  void set_ca_cert_path(const std::string &ca_cert_file_path,
+                        const std::string &ca_cert_dir_path = std::string());
  void set_ca_cert_store(tls::ca_store_t store);
  void load_ca_cert_store(const char *ca_cert, std::size_t size);
  void enable_server_certificate_verification(bool enabled);
+  void enable_server_hostname_verification(bool enabled);
  void enable_system_ca(bool enabled);
 #endif

 private:
  void shutdown_and_close();
-  bool create_stream(std::unique_ptr<Stream> &strm);
+  bool create_stream(std::unique_ptr<Stream> &strm, Error &error,
+                     int &ssl_error, uint64_t &ssl_backend_error);
  void prepare_default_headers(Request &req);

  std::string host_;
@@ -4335,13 +4402,37 @@ private:
  tls::ctx_t tls_ctx_ = nullptr;
  tls::session_t tls_session_ = nullptr;
  std::string ca_cert_file_path_;
+  std::string ca_cert_dir_path_;
  bool custom_ca_loaded_ = false;
  bool certs_loaded_ = false;
  SystemCAMode system_ca_mode_ = SystemCAMode::Auto;
  bool server_certificate_verification_ = true;
+  bool server_hostname_verification_ = true;
 #endif
 };

+template <class Rep, class Period>
+inline void WebSocketClient::set_read_timeout(
+    const std::chrono::duration<Rep, Period> &duration) {
+  detail::duration_to_sec_and_usec(
+      duration, [&](time_t sec, time_t usec) { set_read_timeout(sec, usec); });
+}
+
+template <class Rep, class Period>
+inline void WebSocketClient::set_write_timeout(
+    const std::chrono::duration<Rep, Period> &duration) {
+  detail::duration_to_sec_and_usec(
+      duration, [&](time_t sec, time_t usec) { set_write_timeout(sec, usec); });
+}
+
+template <class Rep, class Period>
+inline void WebSocketClient::set_connection_timeout(
+    const std::chrono::duration<Rep, Period> &duration) {
+  detail::duration_to_sec_and_usec(duration, [&](time_t sec, time_t usec) {
+    set_connection_timeout(sec, usec);
+  });
+}
+
 namespace impl {

 bool is_valid_utf8(const std::string &s);
@@ -1,19 +0,0 @@
-Fix build on NetBSD, which provides posix_spawn_file_actions_addchdir()
-but not the _np() variant.
-
-Upstream PR: https://github.com/sheredom/subprocess.h/pull/102
-Applied locally by scripts/sync_vendor.py until it is merged upstream.
-
-diff --git a/subprocess.h b/subprocess.h
-index 5e809023a4..74a4e006c7 100644
--- a/subprocess.h
-+++ b/subprocess.h
-@@ -1205,7 +1205,7 @@ cleanup:
- 
-   // Set working directory
-   if (process_cwd) {
-#if defined(__APPLE__) && MAC_OS_X_VERSION_MIN_REQUIRED >= 260000
-+#if defined(__NetBSD__) || (defined(__APPLE__) && MAC_OS_X_VERSION_MIN_REQUIRED >= 260000)
-     posix_error = posix_spawn_file_actions_addchdir(&actions, process_cwd);
- #else
- #if defined(__APPLE__) && defined(__clang__)
@@ -1,47 +0,0 @@
-Fix building against glibc older than 2.29, which has no
-posix_spawn_file_actions_addchdir_np (the symbol is genuinely absent from
-libc.so, so no feature-test macro helps). Affects manylinux2014 (glibc 2.17)
-and manylinux_2_28, and was reported on RHEL 8.1. A requested process_cwd now
-fails with ENOSYS there instead of failing the build.
-
-Upstream PR: https://github.com/sheredom/subprocess.h/pull/104
-Applied locally by scripts/sync_vendor.py until it is merged upstream.
-
-(the README.md and test/ changes from the PR are omitted, we only vendor
-subprocess.h; rebased on top of patch-bsd.patch, so apply it after that one)
-
-diff --git a/subprocess.h b/subprocess.h
-index 1ef424a..c363393 100644
--- a/subprocess.h
-+++ b/subprocess.h
-@@ -274,6 +274,21 @@ subprocess_weak int subprocess_alive(struct subprocess_s *const process);
- #include <unistd.h>
- #endif
- 
-+/* Whether subprocess_create_ex can honour process_cwd. glibc only gained
-+   posix_spawn_file_actions_addchdir_np in 2.29. Define this yourself to
-+   override the detection, for instance on musl older than 1.1.24. */
-+#if !defined(SUBPROCESS_HAVE_CWD)
-+#if defined(__GLIBC__)
-+#if __GLIBC_PREREQ(2, 29)
-+#define SUBPROCESS_HAVE_CWD 1
-+#else
-+#define SUBPROCESS_HAVE_CWD 0
-+#endif
-+#else
-+#define SUBPROCESS_HAVE_CWD 1
-+#endif
-+#endif
-+
- #if defined(_WIN32)
- 
- #include <wchar.h>
-@@ -1219,6 +1234,8 @@ cleanup:
-   if (process_cwd) {
- #if defined(__NetBSD__) || (defined(__APPLE__) && MAC_OS_X_VERSION_MIN_REQUIRED >= 260000)
-     posix_error = posix_spawn_file_actions_addchdir(&actions, process_cwd);
-+#elif !SUBPROCESS_HAVE_CWD
-+    posix_error = ENOSYS;
- #else
- #if defined(__APPLE__) && defined(__clang__)
- #pragma clang diagnostic push
@@ -1,107 +0,0 @@
-Fix Windows command line quoting of backslash runs: a trailing backslash, or
-backslashes preceding a double quote, were not doubled, so CommandLineToArgvW
-in the child parsed them as escapes and mangled the argument list.
-
-Upstream PR: https://github.com/sheredom/subprocess.h/pull/101
-Applied locally by scripts/sync_vendor.py until it is merged upstream.
-
-(the test/ changes from the PR are omitted, we only vendor subprocess.h)
-
-diff --git a/subprocess.h b/subprocess.h
-index 5e80902..b06ad4d 100644
--- a/subprocess.h
-+++ b/subprocess.h
-@@ -653,6 +653,7 @@ int subprocess_create_ex(const char *const commandLine[], int options,
-   int wide_len;
-   int i, j;
-   int need_quoting;
-+  subprocess_size_t bs_run;
-   unsigned long flags = 0;
-   unsigned long last_error = 0;
-   int result = subprocess_error_unknown;
-@@ -906,25 +907,29 @@ int subprocess_create_ex(const char *const commandLine[], int options,
-     len++;
- 
-     // Quote the argument if it has a space in it
-    if (strpbrk(commandLine[i], "\t\v ") != SUBPROCESS_NULL ||
-        commandLine[i][0] == SUBPROCESS_NULL)
-+    need_quoting = strpbrk(commandLine[i], "\t\v ") != SUBPROCESS_NULL ||
-+                   commandLine[i][0] == SUBPROCESS_NULL;
-+    if (need_quoting)
-       len += 2;
- 
-+    bs_run = 0;
-     for (j = 0; '\0' != commandLine[i][j]; j++) {
-      switch (commandLine[i][j]) {
-      default:
-        break;
-      case '\\':
-        if (commandLine[i][j + 1] == '"') {
-          len++;
-        }
-+      len++;
- 
-        break;
-      case '"':
-        len++;
-        break;
-+      if ('\\' == commandLine[i][j]) {
-+        bs_run++;
-+      } else {
-+        if ('"' == commandLine[i][j]) {
-+          // Duplicate the preceding run and escape the quote.
-+          len += bs_run + 1;
-+        }
-+        bs_run = 0;
-       }
-      len++;
-+    }
-+
-+    if (need_quoting) {
-+      // Duplicate trailing slashes before the generated closing quote.
-+      len += bs_run;
-     }
-   }
- 
-@@ -949,22 +954,29 @@ int subprocess_create_ex(const char *const commandLine[], int options,
-       commandLineCombined[len++] = '"';
-     }
- 
-    for (j = 0; '\0' != commandLine[i][j]; j++) {
-      switch (commandLine[i][j]) {
-      default:
-        break;
-      case '\\':
-        if (commandLine[i][j + 1] == '"') {
-          commandLineCombined[len++] = '\\';
-        }
-+    for (j = 0; '\0' != commandLine[i][j];) {
-+      bs_run = 0;
-+      while ('\\' == commandLine[i][j]) {
-+        bs_run++;
-+        j++;
-+      }
-+
-+      if ('"' == commandLine[i][j]) {
-+        // 2n + 1 slashes preserve n slashes and escape the quote.
-+        bs_run = (bs_run * 2) + 1;
-+      } else if ('\0' == commandLine[i][j] && need_quoting) {
-+        // 2n slashes preserve n slashes before the closing quote.
-+        bs_run *= 2;
-+      }
- 
-        break;
-      case '"':
-+      while (bs_run > 0) {
-         commandLineCombined[len++] = '\\';
-        break;
-+        bs_run--;
-       }
- 
-      commandLineCombined[len++] = commandLine[i][j];
-+      if ('\0' != commandLine[i][j]) {
-+        commandLineCombined[len++] = commandLine[i][j++];
-+      }
-     }
-     if (need_quoting) {
-       commandLineCombined[len++] = '"';
@@ -107,7 +107,8 @@ enum subprocess_error_e {
  subprocess_error_permission_denied = -5,
  subprocess_error_no_memory = -6,
  subprocess_error_pipe = -7,
-  subprocess_error_spawn = -8
+  subprocess_error_spawn = -8,
+  subprocess_error_not_supported = -9
 };

 #if defined(__cplusplus)
@@ -275,8 +276,10 @@ subprocess_weak int subprocess_alive(struct subprocess_s *const process);
 #endif

 /* Whether subprocess_create_ex can honour process_cwd. glibc only gained
-   posix_spawn_file_actions_addchdir_np in 2.29. Define this yourself to
-   override the detection, for instance on musl older than 1.1.24. */
+   posix_spawn_file_actions_addchdir_np in 2.29, and macOS in 10.15; the SDKs
+   mark it unavailable on iOS, tvOS and watchOS, where the undefined version
+   macro folds to 0 and so answers correctly. Define this yourself to override
+   the detection, for instance on musl older than 1.1.24. */
 #if !defined(SUBPROCESS_HAVE_CWD)
 #if defined(__GLIBC__)
 #if __GLIBC_PREREQ(2, 29)
@@ -284,11 +287,27 @@ subprocess_weak int subprocess_alive(struct subprocess_s *const process);
 #else
 #define SUBPROCESS_HAVE_CWD 0
 #endif
+#elif defined(__APPLE__) && MAC_OS_X_VERSION_MIN_REQUIRED < 101500
+#define SUBPROCESS_HAVE_CWD 0
 #else
 #define SUBPROCESS_HAVE_CWD 1
 #endif
 #endif

+/* Whether posix_spawn reports a failed exec back to the caller. glibc only
+   started doing so in 2.24; before that the child silently exits with 127. */
+#if !defined(SUBPROCESS_SPAWN_REPORTS_EXEC_ERRORS)
+#if defined(__GLIBC__)
+#if __GLIBC_PREREQ(2, 24)
+#define SUBPROCESS_SPAWN_REPORTS_EXEC_ERRORS 1
+#else
+#define SUBPROCESS_SPAWN_REPORTS_EXEC_ERRORS 0
+#endif
+#else
+#define SUBPROCESS_SPAWN_REPORTS_EXEC_ERRORS 1
+#endif
+#endif
+
 #if defined(_WIN32)

 #include <wchar.h>
@@ -554,6 +573,8 @@ int subprocess_error_from_errno(int error) {
  case ENFILE:
  case ENOMEM:
    return subprocess_error_no_memory;
+  case ENOSYS:
+    return subprocess_error_not_supported;
  default:
    return subprocess_error_unknown;
  }
@@ -1358,6 +1379,17 @@ cleanup:
      goto cleanup;
    }
  } else {
+#if !SUBPROCESS_SPAWN_REPORTS_EXEC_ERRORS
+    /* posix_spawn cannot tell us the exec failed, so check up front */
+    if (0 != access(commandLine[0], X_OK)) {
+      saved_errno = errno;
+      result = subprocess_error_from_errno(saved_errno);
+      if (subprocess_error_unknown == result) {
+        result = subprocess_error_spawn;
+      }
+      goto cleanup;
+    }
+#endif
    posix_error = posix_spawn(&child, commandLine[0], &actions,
                              SUBPROCESS_NULL,
                              SUBPROCESS_CONST_CAST(char *const *, commandLine),
Author	SHA1	Message	Date
Hitesh Chopra	d2f83055d6	ggml-cpu : fix CPU affinity mask being ignored on Android (#26838 )	2026-08-10 15:13:40 +03:00
Yash Raj Pandey	f8def7fe16	ggml : require contiguous src for ROLL on CUDA and Metal (#25928 ) ggml_roll only asserts nb[0] == ggml_type_size, so a permuted src is a valid input, but the CUDA and Metal roll kernels index by ne alone and never read the nb strides. A non-contiguous src therefore produced silently wrong results. Neither backend declared a contiguity requirement in supports_op, so the scheduler did not fall back to the CPU implementation, which does handle strides correctly. Add the requirement to both backends, matching the existing GGML_OP_ROPE guard, and add a permuted test_roll case.	2026-08-10 15:01:44 +03:00
Pascal	4dee52f82d	ui: UI/chat form follow ups (#26743 ) * ui: split the markdown rendering setting per surface User content and thinking get their own toggle again, so turning off markdown for a message leaves reasoning blocks formatted. Both default to markdown. A stored renderContentAsRawText unfolds onto the user key and is dropped from the config. File mentions render as badges in the raw text path too, through a narrow pass over [name](file://path) that leaves everything else untouched. * ui: let the rich chat input scroll past its max height The contenteditable renderer caps its height with max-height but had no overflow rule, so a long buffer overflowed into the input area wrapper and got clipped by its overflow-hidden, leaving no way to reach the bottom of the message. The textarea renderer scrolls natively and was never affected. * ui: apply the new lint and format config * ui: move the render keys unfolding into the migration service Address review from @allozaur: the settings store no longer rewrites persisted config on load, the raw text toggle now unfolds onto the per-surface render keys in migration.service.ts, next to the other config migrations. The mention scanner flag and the directory path suffix become named constants.	2026-08-10 13:32:51 +02:00
Sigbjørn Skjæret	e5275f6f77	ci : don't specify python version in server-sanitize for broader runner compatibility (#26840 ) * don't specify python version for broader runner compatibilty * run the workflow	2026-08-10 13:32:22 +02:00
Pascal	4ae84dea27	server: add more tool isolation support (ssh remote + podman rootless) (#26774 ) * server: add an ssh transport to the tools runtime --tools-runtime ssh:<target> runs the built-in tools on a remote host, where target is whatever ssh already resolves, a user@host or a config alias, so no credentials live in llama.cpp. Only build_argv and upload differ from the docker transport: the remote shell re-parses the command line, so the argv travels through shell_quote_join, and files go over scp with the same quoting on the remote path. Authentication is key-based and the host key must already be trusted, since the tools run without a console and any prompt would hang them. The target is validated before use. The spec can reach us from the x-tool-runtime header, and a leading dash would turn it into an ssh option, which is enough to run a command back on the host. Nothing is created and nothing is reclaimed, so an ssh spec goes straight to the tool call instead of through the container runtime. Note that this is remoting rather than isolation: the tools can do whatever the target account can do, and the isolation is whatever runs them on the far side. * server: support podman in the tools runtime docker and podman expose the same run, exec, cp and inspect verbs with the same argument order, so a single implementation drives both and the engine is carried by the spec prefix: podman:<image> and podman-container:<id> sit next to the docker forms. tools_io_docker becomes tools_io_container and the runtime spawner becomes server_tools_container_runtime, both holding the client binary chosen at parse time. A single parse_container_runtime() resolves every spec, so adding another engine is one string in the table. make_tools_io() now rejects the spawning forms. The spec also reaches it from the x-tool-runtime header, which is client controlled, and only the runtime that owns a container is allowed to create one: a tool call can attach to a running container, nothing more. * ./build/bin/llama-gen-docs * server: simplify the tools runtime and drop the file copy step A server_tools_runtime base with one virtual spec() replaces the container runtime and the bare spec string that ssh needed next to it, so server_tools is back to a single pointer and neither setup nor the handler tests which of the two is set. write_file used to spill its content into a temporary file on the host and copy it in, because run_subprocess had no way to feed a child. It now takes an optional stdin payload and creates the parent directory and the file in a single round trip through a shell in the isolate. That removes the upload virtual and both implementations: no more container cp or scp, no second binary on the host, no sftp subsystem on the target, no predictable temporary in a shared tmp, and none of the content reaching an argv the remote shell re-parses. It also fixes write_file over ssh, which never worked: scp speaks sftp and takes the remote path literally, so quoting it kept the quotes in the file name. Writing the payload before reading the output relies on the child draining stdin as it goes, which holds for cat, its only user today. * ./build/bin/llama-gen-docs * server: harden the tools runtime against argv injection and a stdin stall Validate the container id from x-tool-runtime and --tools-runtime the same way the ssh target already is, so an id shaped like an option (docker-container:--privileged) is rejected before it reaches the engine's exec command line instead of running against a hardened container. Feed the child's stdin after the watchdog is armed, so a transport that stalls mid-write is terminated at the deadline rather than blocking the request forever. Cover both guards and fix the unknown-scheme test, which used ssh: as its example and now names a real runtime. * tests: exercise the tools runtime tests on podman as well as docker Follow-up #26507. The container runtime drives docker and podman through one implementation, so parametrize the availability helper, the container fixture and the attach test on the engine, and cover both engine prefixes in the container id injection test. Each engine skips on its own when it is not installed. The spawn cleanup test stays docker only: it recovers the spawned id from the container hostname, which docker sets to the short id and podman rootless does not guarantee. Podman keeps its coverage through the attach path. * server: release the container handle before respawning Follow-up #26507. create() writes over the handle it is given, so a respawn after the container died on its own leaked the pipes and the process handle of the previous one. * server: trim the tools runtime comments * server: read tool output as raw bytes and harden the runtime on Windows The stdout pipe is read with read() instead of fgets(), so a chunk can hold any byte, including NUL, and still streams as soon as data is available. Past the size cap the pipe keeps draining so the child never blocks on a full pipe. Both pipe fds are forced to binary mode on Windows, where the CRT defaults them to text mode and translates line endings in both directions. Stdin is now always closed after the feed: the child reads a deterministic EOF, and the Windows docker and ssh clients stop outliving their command on a stdin pipe that never closes. The attach form of --tools-runtime has no lifecycle to own, so it becomes a static target validated once at startup. This removes the subprocess that ran on every tool call and serialized calls behind a mutex; a stopped container now surfaces the engine's own error at exec time. The cidfile path is passed as UTF-8, matching the encoding the subprocess layer expects for the CreateProcessW command line, so the spawn form works from a non-ASCII Windows profile. The SIGPIPE note in server.cpp now names the tools runtime children as well as the MCP ones. * clean up comments * less pollute global scope * nits * tests: name the container image after both engines --------- Co-authored-by: Xuan Son Nguyen <son@huggingface.co>	2026-08-10 13:31:09 +02:00
Pedro Cuenca	62bf73d25c	model: Muse Glimmer Support (#26841 ) * Get started with Onyx * Add architecture * Skip keys handled in super() * Loading tensors * Shorten * Graph * Apply suggestion from @pcuenca * Remove norm now embedding in transformers weights * Add eot * Explicit output_multiplier * Handle post_norm_eps * No super call; unhardcode eot. The pattern `self._set_vocab_gpt2()` seems preferred throughout the codebase, and it allows `set_vocab()` to be called from a different part of the Python class hierarchy: the drafter model converter that we may need eventually. * Register for drafting * DFlash: inherit rope type from the linked target. Another option would be to store it in the gguf file itself. * mmproj conversion Note: some fields to be renamed after the implementation works. We are keeping compatibility with the reference Meta gguf for testing purposes. * "clip" header declarations * Load mmproj * Pre-processing * Graph * Go back to using delimiters. Otherwise our generations are worse. Transformers does not use them. We need to trace inputs to verify whether they are equivalent. * downsample_factor -> merge_size * Add vision graph lol, forgot from a previous commit * Additional renames, align with llama.cpp / transformers * Prefer _size instead of independent _h and _w * Fix token layout Co-authored-by: Young Han <younghan@fb.com> * onyx: bring the chat parser onto the onyx branch common/chat.cpp on this branch has no Onyx handling, so a converted model serves malformed chat: the assistant preamble leaks into content ("to=self<\|message\|>...") and tool calls fail with HTTP 500 "The model produced output that does not match the expected peg-native format" common_chat_params_init_onyx exists on onyx-fair-patch, added there by 8bb73dd3d. It was never on this branch, so this is not a regression -- the two lines developed independently. The code here is taken verbatim from that commit. It is the clean side of `git merge origin/onyx-fair-patch`: chat.cpp is one of the files that merges without conflict. The full merge is not viable -- it produces 13 conflicts, including add/add on conversion/onyx.py and src/models/onyx.cpp where the q_norm-folding and metadata-scale approaches contradict each other, and #4/#7 are stacked on this branch's side of that. Verified on this branch: builds with 0 errors, converts an Onyx checkpoint, and serving it gives "4" for "What is 2+2?" plus a correct get_weather {"city":"Paris"} tool call, where the unported branch gives the two failures above. No converter or runtime changes are included, so this should not interact with the q_norm work. Co-authored-by: Beto de Paola <betodepaola@meta.com> * Less params, bilinear pos-emb interpolation as a graph op instead of CPU * Map to symbolic V_MMPROJ instead of strings * Make a couple params explicit * Patchify via build_inp() * No param for rope_theta * Small cleanup * Restore blank line * Unpermute, to adapt to the latest transformers checkpoint * Apply norm after token embeddings This follows the latest transformers approach. * Remove duplicated function * build_vit * onyx: use the model rope theta on sliding-window layers * DFlash: conversion from transformers drafter * Revert rope_type derivation from target NOTE: this breaks compatibility with Meta's distributed DFlash GGUFs, as the Q/K are stored in "NEOX" (rotated half) format, like in transformers. * Apply suggestion from @pcuenca * Set model type * Remove comment that will become obsolete * Hardcode post_norm_rms_eps instead of new param * Derive SWA+RoPE pattern from gguf array or scalar * Fix model type <-> number of layers * Reorder * Rename * Fix typo * DFlash: seed the draft KV cache from multimodal embedding batches `common_speculative_impl_draft_dflash::process()` returned early on any batch carrying embeddings, so an image prefill never had its target-layer features fused through the DFlash encoder and injected into the draft's KV cache. That left a hole spanning the image's positions, and the next injection at a post-image position failed to initialize its batch: ``` decoding image batch 1/1, n_tokens_batch = 256 decode: failed to initialize batch llama_decode: failed to decode, ret = -1 process: llama_decode(ctx_dft) failed rc=-1 (n_tokens=17, offset=0) srv decode: failed to process speculative batch ``` Every image request with `--spec-type draft-dflash` failed with HTTP 500. Text-only was unaffected, since those batches carry token ids and were let through. Restore the earlier condition, which admits a batch that is either tokens or embeddings and skips only the degenerate neither/both cases. The rest of `process()` is already layout-agnostic -- it gathers features via `llama_get_embeddings_layer_inp()` and indexes `batch_in.pos[]` / `batch_in.seq_id[]`, none of which assume token ids -- so this is the whole fix. Validated against `muse-glimmer-30B-bf16.gguf` + `mmproj-muse-glimmer-30B-bf16.gguf` + a DFlash draft head, on an image describe-the-shapes request: - before: HTTP 500, `failed to process speculative batch` - after: HTTP 200, draft acceptance 0.34012 (167 accepted / 491 generated), mean len 3.04 Output equivalence holds, which is the property that matters: at temperature 0 the drafted response is byte-identical to the same request served with no draft attached (1213/1213 chars), so the draft is drafting correctly through the image context rather than merely not crashing. * Conversion: prefer rewrite to mapping * Revert "Conversion: prefer rewrite to mapping" This reverts commit `a92d0ac584`. * fix lint * sliding_window metadata is not optional * disable state save/load * Apply suggestion from @pcuenca --------- Co-authored-by: Young Han <younghan@fb.com> Co-authored-by: Beto de Paola <betodepaola@meta.com> Co-authored-by: Daniel Han <michaelhan2050@gmail.com> Co-authored-by: ruanrms <ruanslv@gmail.com> Co-authored-by: Xuan Son Nguyen <son@huggingface.co> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co>	2026-08-10 13:07:27 +02:00
Guido Imperiale	a52077c4ca	chat : Align Laguna-S-2.1 chat template to huggingface (#26232 )	2026-08-10 05:20:59 -05:00
Pascal	4c6766fd7e	vendor: sync subprocess.h and drop local patches (#26808 ) Upstream merged the Windows argument quoting fix, the NetBSD build fix and the chdir fallback for glibc older than 2.29, so pin the vendored copy to a commit that carries all three and remove the patch files along with the apply step in the sync script. The new pin also brings the exec error report on glibc older than 2.24 and the ENOSYS mapping to a dedicated error code. Both are additive and no caller inspects those values.	2026-08-10 11:59:08 +02:00
Pedro Cuenca	86c298fb8a	llama: Restore quantization of mmprojs (#26818 ) * Restore quantization of mmprojs This was lost in the refactor undertaken in #22004. * add noreturn --------- Co-authored-by: Xuan Son Nguyen <son@huggingface.co>	2026-08-10 11:58:32 +02:00
shivamkumard-ctrl	2e2d99cfd2	ci: Add support for CUDA 13.4 ARM64 builds for Windows (#26650 ) * ci: Add support for CUDA 13.4 ARM64 builds for Windows Added an architecture-specific CUDA 13.4 Windows build entry targeting ARM64. Added a CMake configuration to enable ARM64 CUDA cross-compilation from an x64 Windows environment using the x64-hosted CUDA and MSVC toolchain while linking against the ARM64 CUDA import libraries to produce ggml-cuda.dll. Validated the self-hosted Windows x64 workflow, including toolkit acquisition, CMake configuration, ARM64 CUDA cross-compilation, and packaging. Runtime validation was performed separately on a native ARM64 RTX Spark system using TinyLlama 1.1B Q4_K_M to verify the generated binaries. The ARM64 CUDA job builds only the ggml-cuda.dll backend (LLAMA_BUILD_SERVER=OFF). The release consists of two packages: the main ARM64 release package, which combines the existing ARM64 CPU outputs with ggml-cuda.dll, and a separate runtime package containing the required CUDA runtime libraries (cudart64_13.dll, cublas64_13.dll, and cublasLt64_13.dll). The CUDA 13.4 setup uses NVIDIA Developer Preview component archives instead of the GA component downloads used by the existing CUDA setups and will require updates once CUDA 13.4 reaches GA. * ci: cleans up to align with x64 CUDA setup - Moves CUDA-specific CMake options into matrix defines. - Keeps the CUB 3DOT2 option only for CUDA 12.4. - Removes runtime argument construction and the unnecessary server option. - Aligns ARM64 CUDA runtime packaging with the existing robocopy approach. - Generalizes the ARM64 release label from CUDA 13.4 to CUDA 13. * ci: Set CUDA job name as version-architecture pair * mark as preview Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2026-08-10 11:46:44 +03:00
Ruixiang Wang	7a20b417f4	model: add MTP support for Nemotron model (#26725 ) * model: add MTP support for Nemotron Nano model * model: add mtp_flags for nemotron model * address review comments	2026-08-10 11:25:24 +03:00
Alessandro de Oliveira Faria (A.K.A.CABELO)	e23e9440eb	vendor : update cpp-httplib to 0.53.0 (#26821 )	2026-08-10 09:57:45 +02:00