Merge branch 'upstream' into concedo_experimental

# Conflicts:
#	.github/workflows/build.yml
#	.github/workflows/release.yml
#	.github/workflows/winget.yml
#	CMakeLists.txt
#	CODEOWNERS
#	CONTRIBUTING.md
#	cmake/build-info.cmake
#	docs/ops.md
#	docs/ops/BLAS.csv
#	docs/ops/Metal.csv
#	examples/CMakeLists.txt
#	examples/save-load-state/save-load-state.cpp
#	examples/simple-cmake-pkg/README.md
#	ggml/CMakeLists.txt
#	ggml/src/CMakeLists.txt
#	ggml/src/ggml-rpc/ggml-rpc.cpp
#	ggml/src/ggml-webgpu/ggml-webgpu.cpp
#	ggml/src/ggml-webgpu/wgsl-shaders/embed_wgsl.py
#	src/llama-quant.cpp
#	tests/test-backend-ops.cpp
#	tools/server/CMakeLists.txt

tools/server/server-context.cpp

@@ -101,8 +101,6 @@ struct server_slot {
     std::string  generated_text;
     llama_tokens generated_tokens;
 
-    common_chat_msg chat_msg;
-
     std::vector<completion_token_output> generated_token_probs;
 
     bool has_next_token = true;
@@ -153,9 +151,6 @@ struct server_slot {
 
     llama_token sampled;
 
-    common_chat_format chat_format = COMMON_CHAT_FORMAT_CONTENT_ONLY;
-    std::vector<std::string> generated_tool_call_ids;
-
     // stats
     size_t n_sent_text = 0; // number of sent text character
 
@@ -183,13 +178,10 @@ struct server_slot {
         stop           = STOP_TYPE_NONE;
         stopping_word  = "";
         n_sent_text    = 0;
-        chat_format    = COMMON_CHAT_FORMAT_CONTENT_ONLY;
 
         generated_tokens.clear();
         generated_token_probs.clear();
-        chat_msg = {};
         json_schema = json();
-        generated_tool_call_ids.clear();
 
         // clear speculative decoding stats
         n_draft_total = 0;
@@ -302,23 +294,6 @@ struct server_slot {
         return timings;
     }
 
-    const common_chat_msg & update_chat_msg(std::vector<common_chat_msg_diff> & diffs) {
-        GGML_ASSERT(task);
-
-        auto previous_msg = chat_msg;
-        SRV_DBG("Parsing chat message: %s\n", generated_text.c_str());
-        auto new_msg = common_chat_parse(
-            generated_text,
-            /* is_partial= */ stop != STOP_TYPE_EOS,
-            task->params.oaicompat_chat_syntax);
-        if (!new_msg.empty()) {
-            new_msg.set_tool_call_ids(generated_tool_call_ids, gen_tool_call_id);
-            chat_msg = new_msg;
-            diffs = common_chat_msg_diff::compute_diffs(previous_msg, new_msg.empty() ? previous_msg : new_msg);
-        }
-        return chat_msg;
-    }
-
     size_t find_stopping_strings(const std::string & text, const size_t last_token_size, bool is_full_stop) {
         GGML_ASSERT(task);
 
@@ -1284,8 +1259,6 @@ struct server_context_impl {
         } else {
             res->content = tkn.text_to_send;
             res->tokens  = { tkn.tok };
-
-            slot.update_chat_msg(res->oaicompat_msg_diffs);
         }
 
         res->n_decoded           = slot.n_decoded;
@@ -1317,8 +1290,14 @@ struct server_context_impl {
         res->id_slot = slot.id;
 
         res->index           = slot.task->index;
-        res->content         = slot.generated_text;
-        res->tokens          = std::move(slot.generated_tokens);
+        // in stream mode, content and tokens are already in last partial chunk
+        if (slot.task->params.stream) {
+            res->content     = "";
+            res->tokens      = llama_tokens{};
+        } else {
+            res->content     = std::move(slot.generated_text);
+            res->tokens      = std::move(slot.generated_tokens);
+        }
         res->timings         = slot.get_timings();
         res->prompt          = slot.task->tokens.detokenize(ctx, true);
         res->response_fields = std::move(slot.task->params.response_fields);
@@ -1338,7 +1317,6 @@ struct server_context_impl {
         res->res_type          = slot.task->params.res_type;
         res->oaicompat_model   = slot.task->params.oaicompat_model;
         res->oaicompat_cmpl_id = slot.task->params.oaicompat_cmpl_id;
-        res->oaicompat_msg     = slot.update_chat_msg(res->oaicompat_msg_diffs);
 
         // populate res.probs_output
         if (slot.task->params.sampling.n_probs > 0) {
@@ -2596,6 +2574,9 @@ static std::unique_ptr<server_res_generator> handle_completions_impl(
     try {
         std::vector<server_task> tasks;
 
+        // tracking generation state and partial tool calls
+        std::vector<task_result_state> states;
+
         const auto & prompt = data.at("prompt");
         // TODO: this log can become very long, put it behind a flag or think about a more compact format
         //SRV_DBG("Prompt: %s\n", prompt.is_string() ? prompt.get<std::string>().c_str() : prompt.dump(2).c_str());
@@ -2611,6 +2592,7 @@ static std::unique_ptr<server_res_generator> handle_completions_impl(
             inputs = tokenize_input_prompts(ctx_server.vocab, ctx_server.mctx, prompt, true, true);
         }
         tasks.reserve(inputs.size());
+        states.reserve(inputs.size());
         for (size_t i = 0; i < inputs.size(); i++) {
             server_task task = server_task(type);
 
@@ -2628,10 +2610,12 @@ static std::unique_ptr<server_res_generator> handle_completions_impl(
             task.params.res_type          = res_type;
             task.params.oaicompat_cmpl_id = completion_id;
             task.params.oaicompat_model   = ctx_server.model_name;
+            states.push_back(task.params.oaicompat_chat_syntax);
 
             tasks.push_back(std::move(task));
         }
 
+        rd.set_states(std::move(states));
         rd.post_tasks(std::move(tasks));
     } catch (const std::exception & e) {
         res->error(format_error_response(e.what(), ERROR_TYPE_INVALID_REQUEST));
@@ -2657,7 +2641,6 @@ static std::unique_ptr<server_res_generator> handle_completions_impl(
             // if single request, return single object instead of array
             res->ok(arr.size() == 1 ? arr[0] : arr);
         }
-
     } else {
         // in streaming mode, the first error must be treated as non-stream response
         // this is to match the OAI API behavior
@@ -2676,76 +2659,92 @@ static std::unique_ptr<server_res_generator> handle_completions_impl(
         }
 
         // next responses are streamed
+        // to be sent immediately
+        json first_result_json = first_result->to_json();
         if (res_type == TASK_RESPONSE_TYPE_ANTHROPIC) {
-            res->data = format_anthropic_sse(first_result->to_json());
+            res->data = format_anthropic_sse(first_result_json);
         } else {
-            res->data = format_oai_sse(first_result->to_json()); // to be sent immediately
+            res->data = format_oai_sse(first_result_json);
         }
         res->status = 200;
         res->content_type = "text/event-stream";
         res->next = [res_this = res.get(), res_type, &should_stop](std::string & output) -> bool {
-            if (should_stop()) {
-                SRV_DBG("%s", "stopping streaming due to should_stop condition\n");
-                return false; // should_stop condition met
-            }
-
-            if (!res_this->data.empty()) {
-                // flush the first chunk
-                output = std::move(res_this->data);
-                res_this->data.clear();
-                return true;
-            }
-
-            server_response_reader & rd = res_this->rd;
-
-            // check if there is more data
-            if (!rd.has_next()) {
+            static auto format_error = [](task_response_type res_type, const json & res_json) {
                 if (res_type == TASK_RESPONSE_TYPE_ANTHROPIC) {
-                    // Anthropic doesn't send [DONE], message_stop was already sent
-                    output = "";
-                } else if (res_type != TASK_RESPONSE_TYPE_NONE) {
-                    output = "data: [DONE]\n\n";
-                } else {
-                    output = "";
-                }
-                SRV_DBG("%s", "all results received, terminating stream\n");
-                return false; // no more data, terminate
-            }
-
-            // receive subsequent results
-            auto result = rd.next(should_stop);
-            if (result == nullptr) {
-                SRV_DBG("%s", "stopping streaming due to should_stop condition\n");
-                return false; // should_stop condition met
-            }
-
-            // send the results
-            json res_json = result->to_json();
-            if (result->is_error()) {
-                if (res_type == TASK_RESPONSE_TYPE_ANTHROPIC) {
-                    output = format_anthropic_sse({
+                    return format_anthropic_sse({
                         {"event", "error"},
                         {"data", res_json},
                     });
                 } else {
-                    output = format_oai_sse(json {{ "error", res_json }});
+                    return format_oai_sse(json {{ "error", res_json }});
                 }
-                SRV_DBG("%s", "error received during streaming, terminating stream\n");
-                return false; // terminate on error
-            } else {
-                GGML_ASSERT(
-                    dynamic_cast<server_task_result_cmpl_partial*>(result.get()) != nullptr
-                    || dynamic_cast<server_task_result_cmpl_final*>(result.get()) != nullptr
-                );
-                if (res_type == TASK_RESPONSE_TYPE_ANTHROPIC) {
-                    output = format_anthropic_sse(res_json);
-                } else {
-                    output = format_oai_sse(res_json);
-                }
-            }
+            };
 
-            // has next data, continue
-            return true;
+            try {
+                if (should_stop()) {
+                    SRV_DBG("%s", "stopping streaming due to should_stop condition\n");
+                    return false; // should_stop condition met
+                }
+
+                if (!res_this->data.empty()) {
+                    // flush the first chunk
+                    output = std::move(res_this->data);
+                    res_this->data.clear();
+                    return true;
+                }
+
+                server_response_reader & rd = res_this->rd;
+
+                // check if there is more data
+                if (!rd.has_next()) {
+                    if (res_type == TASK_RESPONSE_TYPE_ANTHROPIC) {
+                        // Anthropic doesn't send [DONE], message_stop was already sent
+                        output = "";
+                    } else if (res_type != TASK_RESPONSE_TYPE_NONE) {
+                        output = "data: [DONE]\n\n";
+                    } else {
+                        output = "";
+                    }
+                    SRV_DBG("%s", "all results received, terminating stream\n");
+                    return false; // no more data, terminate
+                }
+
+                // receive subsequent results
+                auto result = rd.next(should_stop);
+                if (result == nullptr) {
+                    SRV_DBG("%s", "stopping streaming due to should_stop condition\n");
+                    return false; // should_stop condition met
+                }
+
+                // send the results
+                if (result->is_error()) {
+                    json res_json = result->to_json();
+                    output = format_error(res_type, res_json);
+                    SRV_DBG("%s", "error received during streaming, terminating stream\n");
+                    return false; // terminate on error
+                } else {
+                    GGML_ASSERT(
+                        dynamic_cast<server_task_result_cmpl_partial*>(result.get()) != nullptr
+                        || dynamic_cast<server_task_result_cmpl_final*>(result.get()) != nullptr
+                    );
+                    json res_json = result->to_json();
+                    if (res_type == TASK_RESPONSE_TYPE_ANTHROPIC) {
+                        output = format_anthropic_sse(res_json);
+                    } else {
+                        output = format_oai_sse(res_json);
+                    }
+                }
+
+                // has next data, continue
+                return true;
+
+            } catch (const std::exception & e) {
+                json error_json = format_error_response(e.what(), ERROR_TYPE_SERVER);
+                output = format_error(res_type, error_json);
+
+                // terminate on exception
+                return false;
+            }
         };
     }
 

tools/server/server-models.cpp

@@ -900,6 +900,7 @@ static bool should_strip_proxy_header(const std::string & header_name) {
     // Headers that get duplicated when router forwards child responses
     if (header_name == "server" ||
         header_name == "transfer-encoding" ||
+        header_name == "content-length" || // quick fix for https://github.com/ggml-org/llama.cpp/issues/17710
         header_name == "keep-alive") {
         return true;
     }

tools/server/server-queue.cpp

@@ -271,6 +271,10 @@ void server_response::terminate() {
 // server_response_reader
 //
 
+void server_response_reader::set_states(std::vector<task_result_state> && states) {
+    this->states = std::move(states);
+}
+
 void server_response_reader::post_tasks(std::vector<server_task> && tasks) {
     id_tasks = server_task::get_list_id(tasks);
     queue_results.add_waiting_tasks(tasks);
@@ -298,6 +302,12 @@ server_task_result_ptr server_response_reader::next(const std::function<bool()>
                 SRV_DBG("%s", "received error result, stopping further processing\n");
                 return result;
             }
+            if (!states.empty()) {
+                // update the generation state if needed
+                size_t idx = result->get_index();
+                GGML_ASSERT(idx < states.size());
+                result->update(states[idx]);
+            }
             if (result->is_stop()) {
                 received_count++;
             }

tools/server/server-queue.h

@@ -7,6 +7,8 @@
 #include <mutex>
 #include <unordered_set>
 
+// struct for managing server tasks
+// in most cases, use server_response_reader to post new tasks and retrieve results
 struct server_queue {
 private:
     int id = 0;
@@ -67,6 +69,8 @@ private:
     void cleanup_pending_task(int id_target);
 };
 
+// struct for managing server responses
+// in most cases, use server_response_reader to retrieve results
 struct server_response {
 private:
     bool running = true;
@@ -120,6 +124,10 @@ struct server_response_reader {
     bool cancelled = false;
     int polling_interval_seconds;
 
+    // tracking generation state and partial tool calls
+    // only used by streaming completions
+    std::vector<task_result_state> states;
+
     // should_stop function will be called each polling_interval_seconds
     server_response_reader(std::pair<server_queue &, server_response &> server_queues, int polling_interval_seconds)
         : queue_tasks(server_queues.first), queue_results(server_queues.second), polling_interval_seconds(polling_interval_seconds) {}
@@ -127,6 +135,7 @@ struct server_response_reader {
         stop();
     }
 
+    void set_states(std::vector<task_result_state> && states);
     void post_tasks(std::vector<server_task> && tasks);
     bool has_next() const;
 

tools/server/server-task.cpp

@@ -565,6 +565,7 @@ std::vector<unsigned char> completion_token_output::str_to_bytes(const std::stri
 // server_task_result_cmpl_final
 //
 json server_task_result_cmpl_final::to_json() {
+    GGML_ASSERT(is_updated && "update() must be called before to_json()");
     switch (res_type) {
         case TASK_RESPONSE_TYPE_NONE:
             return to_json_non_oaicompat();
@@ -582,8 +583,8 @@ json server_task_result_cmpl_final::to_json() {
 json server_task_result_cmpl_final::to_json_non_oaicompat() {
     json res = json {
         {"index",               index},
-        {"content",             stream ? "" : content}, // in stream mode, content is already in last partial chunk
-        {"tokens",              stream ? llama_tokens {} : tokens},
+        {"content",             content},
+        {"tokens",              tokens},
         {"id_slot",             id_slot},
         {"stop",                true},
         {"model",               oaicompat_model},
@@ -619,7 +620,7 @@ json server_task_result_cmpl_final::to_json_oaicompat() {
     json res = json {
         {"choices",            json::array({
             json{
-                {"text",          stream ? "" : content}, // in stream mode, content is already in last partial chunk
+                {"text",          content},
                 {"index",         index},
                 {"logprobs",      logprobs},
                 {"finish_reason", finish_reason},
@@ -700,6 +701,25 @@ json server_task_result_cmpl_final::to_json_oaicompat_chat() {
     return res;
 }
 
+common_chat_msg task_result_state::update_chat_msg(
+        const std::string & text_added,
+        bool is_partial,
+        std::vector<common_chat_msg_diff> & diffs) {
+    generated_text += text_added;
+    auto msg_prv_copy = chat_msg;
+    SRV_DBG("Parsing chat message: %s\n", generated_text.c_str());
+    auto new_msg = common_chat_parse(
+        generated_text,
+        is_partial,
+        oaicompat_chat_syntax);
+    if (!new_msg.empty()) {
+        new_msg.set_tool_call_ids(generated_tool_call_ids, gen_tool_call_id);
+        chat_msg = new_msg;
+        diffs = common_chat_msg_diff::compute_diffs(msg_prv_copy, new_msg.empty() ? msg_prv_copy : new_msg);
+    }
+    return chat_msg;
+}
+
 json server_task_result_cmpl_final::to_json_oaicompat_chat_stream() {
     std::time_t t = std::time(0);
     std::string finish_reason = "length";
@@ -956,6 +976,7 @@ json server_task_result_cmpl_final::to_json_anthropic_stream() {
 // server_task_result_cmpl_partial
 //
 json server_task_result_cmpl_partial::to_json() {
+    GGML_ASSERT(is_updated && "update() must be called before to_json()");
     switch (res_type) {
         case TASK_RESPONSE_TYPE_NONE:
             return to_json_non_oaicompat();

tools/server/server-task.h

@@ -161,6 +161,25 @@ struct result_prompt_progress {
     json to_json() const;
 };
 
+// struct for tracking the state of a task (e.g., for streaming)
+struct task_result_state {
+    // tracking diffs for partial tool calls
+    std::vector<common_chat_msg_diff> diffs;
+    common_chat_syntax oaicompat_chat_syntax;
+    common_chat_msg chat_msg;
+    std::string generated_text; // append new chunks of generated text here
+    std::vector<std::string> generated_tool_call_ids;
+
+    task_result_state(const common_chat_syntax & oaicompat_chat_syntax)
+        : oaicompat_chat_syntax(oaicompat_chat_syntax) {}
+
+    // parse partial tool calls and update the internal state
+    common_chat_msg update_chat_msg(
+        const std::string & text_added,
+        bool is_partial,
+        std::vector<common_chat_msg_diff> & diffs);
+};
+
 struct server_task_result {
     int id           = -1;
     int id_slot      = -1;
@@ -175,6 +194,9 @@ struct server_task_result {
     virtual int get_index() {
         return -1;
     }
+    virtual void update(task_result_state &) {
+        // only used by server_task_result_cmpl_*
+    }
     virtual json to_json() = 0;
     virtual ~server_task_result() = default;
 };
@@ -233,9 +255,10 @@ struct server_task_result_cmpl_final : server_task_result {
     task_response_type res_type = TASK_RESPONSE_TYPE_NONE;
     std::string        oaicompat_model;
     std::string        oaicompat_cmpl_id;
-    common_chat_msg    oaicompat_msg;
+    common_chat_msg    oaicompat_msg; // to be populated by update()
 
-    std::vector<common_chat_msg_diff> oaicompat_msg_diffs;
+    std::vector<common_chat_msg_diff> oaicompat_msg_diffs; // to be populated by update()
+    bool is_updated = false;
 
     virtual int get_index() override {
         return index;
@@ -247,6 +270,11 @@ struct server_task_result_cmpl_final : server_task_result {
 
     virtual json to_json() override;
 
+    virtual void update(task_result_state & state) override {
+        is_updated = true;
+        oaicompat_msg = state.update_chat_msg(content, false, oaicompat_msg_diffs);
+    }
+
     json to_json_non_oaicompat();
 
     json to_json_oaicompat();
@@ -280,7 +308,8 @@ struct server_task_result_cmpl_partial : server_task_result {
     task_response_type res_type = TASK_RESPONSE_TYPE_NONE;
     std::string        oaicompat_model;
     std::string        oaicompat_cmpl_id;
-    std::vector<common_chat_msg_diff> oaicompat_msg_diffs;
+    std::vector<common_chat_msg_diff> oaicompat_msg_diffs; // to be populated by update()
+    bool is_updated = false;
 
     virtual int get_index() override {
         return index;
@@ -292,6 +321,11 @@ struct server_task_result_cmpl_partial : server_task_result {
 
     virtual json to_json() override;
 
+    virtual void update(task_result_state & state) override {
+        is_updated = true;
+        state.update_chat_msg(content, true, oaicompat_msg_diffs);
+    }
+
     json to_json_non_oaicompat();
 
     json to_json_oaicompat();

tools/server/webui/docs/flows/data-flow-simplified-router-mode.md

@@ -15,7 +15,7 @@ sequenceDiagram
     Stores->>DB: load conversations
     Stores->>API: GET /props
     API-->>Stores: {role: "router"}
-    Stores->>API: GET /models
+    Stores->>API: GET /v1/models
     API-->>Stores: models[] with status (loaded/available)
     loop each loaded model
         Stores->>API: GET /props?model=X
@@ -28,7 +28,7 @@ sequenceDiagram
     alt model not loaded
         Stores->>API: POST /models/load
         loop poll status
-            Stores->>API: GET /models
+            Stores->>API: GET /v1/models
             API-->>Stores: check if loaded
         end
         Stores->>API: GET /props?model=X

tools/server/webui/docs/flows/models-flow.md

@@ -56,7 +56,7 @@ sequenceDiagram
     UI->>modelsStore: fetchRouterModels()
     activate modelsStore
     modelsStore->>ModelsSvc: listRouter()
-    ModelsSvc->>API: GET /models
+    ModelsSvc->>API: GET /v1/models
     API-->>ModelsSvc: ApiRouterModelsListResponse
     Note right of API: {data: [{id, status, path, in_cache}]}
     modelsStore->>modelsStore: routerModels = $state(data)
@@ -132,7 +132,7 @@ sequenceDiagram
     loop poll every 500ms (max 60 attempts)
         modelsStore->>modelsStore: fetchRouterModels()
         modelsStore->>ModelsSvc: listRouter()
-        ModelsSvc->>API: GET /models
+        ModelsSvc->>API: GET /v1/models
         API-->>ModelsSvc: models[]
         modelsStore->>modelsStore: getModelStatus(modelId)
         alt status === LOADED
@@ -165,7 +165,7 @@ sequenceDiagram
     modelsStore->>modelsStore: pollForModelStatus(modelId, UNLOADED)
     loop poll until unloaded
         modelsStore->>ModelsSvc: listRouter()
-        ModelsSvc->>API: GET /models
+        ModelsSvc->>API: GET /v1/models
     end
 
     modelsStore->>modelsStore: modelLoadingStates.set(modelId, false)

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatForm.svelte

@@ -64,7 +64,10 @@
 	let fileInputRef: ChatFormFileInputInvisible | undefined = $state(undefined);
 	let isRecording = $state(false);
 	let message = $state('');
-	let pasteLongTextToFileLength = $derived(Number(currentConfig.pasteLongTextToFileLen) || 2500);
+	let pasteLongTextToFileLength = $derived.by(() => {
+		const n = Number(currentConfig.pasteLongTextToFileLen);
+		return Number.isNaN(n) ? 2500 : n;
+	});
 	let previousIsLoading = $state(isLoading);
 	let recordingSupported = $state(false);
 	let textareaRef: ChatFormTextarea | undefined = $state(undefined);

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessages.svelte

@@ -1,6 +1,5 @@
 <script lang="ts">
 	import { ChatMessage } from '$lib/components/app';
-	import { DatabaseService } from '$lib/services/database';
 	import { chatStore } from '$lib/stores/chat.svelte';
 	import { conversationsStore, activeConversation } from '$lib/stores/conversations.svelte';
 	import { getMessageSiblings } from '$lib/utils';
@@ -19,7 +18,7 @@
 		const conversation = activeConversation();
 
 		if (conversation) {
-			DatabaseService.getConversationMessages(conversation.id).then((messages) => {
+			conversationsStore.getConversationMessages(conversation.id).then((messages) => {
 				allConversationMessages = messages;
 			});
 		} else {

tools/server/webui/src/lib/components/app/chat/ChatSettings/ChatSettingsFields.svelte

@@ -7,7 +7,6 @@
 	import { Textarea } from '$lib/components/ui/textarea';
 	import { SETTING_CONFIG_DEFAULT, SETTING_CONFIG_INFO } from '$lib/constants/settings-config';
 	import { settingsStore } from '$lib/stores/settings.svelte';
-	import { ParameterSyncService } from '$lib/services/parameter-sync';
 	import { ChatSettingsParameterSourceIndicator } from '$lib/components/app';
 	import type { Component } from 'svelte';
 
@@ -22,7 +21,7 @@
 
 	// Helper function to get parameter source info for syncable parameters
 	function getParameterSourceInfo(key: string) {
-		if (!ParameterSyncService.canSyncParameter(key)) {
+		if (!settingsStore.canSyncParameter(key)) {
 			return null;
 		}
 

tools/server/webui/src/lib/components/app/chat/ChatSettings/ChatSettingsImportExportTab.svelte

@@ -2,9 +2,8 @@
 	import { Download, Upload } from '@lucide/svelte';
 	import { Button } from '$lib/components/ui/button';
 	import { DialogConversationSelection } from '$lib/components/app';
-	import { DatabaseService } from '$lib/services/database';
 	import { createMessageCountMap } from '$lib/utils';
-	import { conversationsStore } from '$lib/stores/conversations.svelte';
+	import { conversationsStore, conversations } from '$lib/stores/conversations.svelte';
 
 	let exportedConversations = $state<DatabaseConversation[]>([]);
 	let importedConversations = $state<DatabaseConversation[]>([]);
@@ -21,15 +20,15 @@
 
 	async function handleExportClick() {
 		try {
-			const allConversations = await DatabaseService.getAllConversations();
+			const allConversations = conversations();
 			if (allConversations.length === 0) {
 				alert('No conversations to export');
 				return;
 			}
 
 			const conversationsWithMessages = await Promise.all(
-				allConversations.map(async (conv) => {
-					const messages = await DatabaseService.getConversationMessages(conv.id);
+				allConversations.map(async (conv: DatabaseConversation) => {
+					const messages = await conversationsStore.getConversationMessages(conv.id);
 					return { conv, messages };
 				})
 			);
@@ -47,7 +46,7 @@
 		try {
 			const allData: ExportedConversations = await Promise.all(
 				selectedConversations.map(async (conv) => {
-					const messages = await DatabaseService.getConversationMessages(conv.id);
+					const messages = await conversationsStore.getConversationMessages(conv.id);
 					return { conv: $state.snapshot(conv), messages: $state.snapshot(messages) };
 				})
 			);
@@ -135,9 +134,7 @@
 				.snapshot(fullImportData)
 				.filter((item) => selectedIds.has(item.conv.id));
 
-			await DatabaseService.importConversations(selectedData);
-
-			await conversationsStore.loadConversations();
+			await conversationsStore.importConversationsData(selectedData);
 
 			importedConversations = selectedConversations;
 			showImportSummary = true;

tools/server/webui/src/lib/components/app/dialogs/DialogModelInformation.svelte

@@ -3,8 +3,7 @@
 	import * as Table from '$lib/components/ui/table';
 	import { BadgeModality, CopyToClipboardIcon } from '$lib/components/app';
 	import { serverStore } from '$lib/stores/server.svelte';
-	import { modelsStore } from '$lib/stores/models.svelte';
-	import { ChatService } from '$lib/services/chat';
+	import { modelsStore, modelOptions, modelsLoading } from '$lib/stores/models.svelte';
 	import { formatFileSize, formatParameters, formatNumber } from '$lib/utils';
 
 	interface Props {
@@ -16,38 +15,24 @@
 
 	let serverProps = $derived(serverStore.props);
 	let modelName = $derived(modelsStore.singleModelName);
+	let models = $derived(modelOptions());
+	let isLoadingModels = $derived(modelsLoading());
+
+	// Get the first model for single-model mode display
+	let firstModel = $derived(models[0] ?? null);
 
 	// Get modalities from modelStore using the model ID from the first model
-	// For now it supports only for single-model mode, will be extended with further improvements for multi-model functioanlities
 	let modalities = $derived.by(() => {
-		if (!modelsData?.data?.[0]?.id) return [];
-
-		return modelsStore.getModelModalitiesArray(modelsData.data[0].id);
+		if (!firstModel?.id) return [];
+		return modelsStore.getModelModalitiesArray(firstModel.id);
 	});
 
-	let modelsData = $state<ApiModelListResponse | null>(null);
-	let isLoadingModels = $state(false);
-
-	// Fetch models data when dialog opens
+	// Ensure models are fetched when dialog opens
 	$effect(() => {
-		if (open && !modelsData) {
-			loadModelsData();
+		if (open && models.length === 0) {
+			modelsStore.fetch();
 		}
 	});
-
-	async function loadModelsData() {
-		isLoadingModels = true;
-
-		try {
-			modelsData = await ChatService.getModels();
-		} catch (error) {
-			console.error('Failed to load models data:', error);
-			// Set empty data to prevent infinite loading
-			modelsData = { object: 'list', data: [] };
-		} finally {
-			isLoadingModels = false;
-		}
-	}
 </script>
 
 <Dialog.Root bind:open {onOpenChange}>
@@ -70,8 +55,8 @@
 				<div class="flex items-center justify-center py-8">
 					<div class="text-sm text-muted-foreground">Loading model information...</div>
 				</div>
-			{:else if modelsData && modelsData.data.length > 0}
-				{@const modelMeta = modelsData.data[0].meta}
+			{:else if firstModel}
+				{@const modelMeta = firstModel.meta}
 
 				{#if serverProps}
 					<Table.Root>

tools/server/webui/src/lib/constants/supported-file-types.ts

@@ -126,8 +126,13 @@ export const TEXT_FILE_TYPES = {
 		mimeTypes: [MimeTypeText.JAVA]
 	},
 	[FileTypeText.CPP]: {
-		extensions: [FileExtensionText.CPP, FileExtensionText.C, FileExtensionText.H],
-		mimeTypes: [MimeTypeText.CPP_SRC, MimeTypeText.C_SRC, MimeTypeText.C_HDR]
+		extensions: [
+			FileExtensionText.CPP,
+			FileExtensionText.C,
+			FileExtensionText.H,
+			FileExtensionText.HPP
+		],
+		mimeTypes: [MimeTypeText.CPP_SRC, MimeTypeText.CPP_HDR, MimeTypeText.C_SRC, MimeTypeText.C_HDR]
 	},
 	[FileTypeText.PHP]: {
 		extensions: [FileExtensionText.PHP],
@@ -183,10 +188,30 @@ export const TEXT_FILE_TYPES = {
 	},
 	[FileTypeText.LATEX]: {
 		extensions: [FileExtensionText.TEX],
-		mimeTypes: [MimeTypeText.LATEX]
+		mimeTypes: [MimeTypeText.LATEX, MimeTypeText.TEX, MimeTypeText.TEX_APP]
 	},
 	[FileTypeText.BIBTEX]: {
 		extensions: [FileExtensionText.BIB],
 		mimeTypes: [MimeTypeText.BIBTEX]
+	},
+	[FileTypeText.CUDA]: {
+		extensions: [FileExtensionText.CU, FileExtensionText.CUH],
+		mimeTypes: [MimeTypeText.CUDA]
+	},
+	[FileTypeText.VULKAN]: {
+		extensions: [FileExtensionText.COMP],
+		mimeTypes: [MimeTypeText.PLAIN]
+	},
+	[FileTypeText.HASKELL]: {
+		extensions: [FileExtensionText.HS],
+		mimeTypes: [MimeTypeText.HASKELL]
+	},
+	[FileTypeText.CSHARP]: {
+		extensions: [FileExtensionText.CS],
+		mimeTypes: [MimeTypeText.CSHARP]
+	},
+	[FileTypeText.PROPERTIES]: {
+		extensions: [FileExtensionText.PROPERTIES],
+		mimeTypes: [MimeTypeText.PROPERTIES]
 	}
 } as const;

tools/server/webui/src/lib/enums/files.ts

@@ -62,7 +62,12 @@ export enum FileTypeText {
 	VUE = 'vue',
 	SVELTE = 'svelte',
 	LATEX = 'latex',
-	BIBTEX = 'bibtex'
+	BIBTEX = 'bibtex',
+	CUDA = 'cuda',
+	VULKAN = 'vulkan',
+	HASKELL = 'haskell',
+	CSHARP = 'csharp',
+	PROPERTIES = 'properties'
 }
 
 // File extension enums
@@ -121,7 +126,14 @@ export enum FileExtensionText {
 	VUE = '.vue',
 	SVELTE = '.svelte',
 	TEX = '.tex',
-	BIB = '.bib'
+	BIB = '.bib',
+	CU = '.cu',
+	CUH = '.cuh',
+	COMP = '.comp',
+	HPP = '.hpp',
+	HS = '.hs',
+	PROPERTIES = '.properties',
+	CS = '.cs'
 }
 
 // MIME type enums
@@ -165,7 +177,10 @@ export enum MimeTypeText {
 	CSV = 'text/csv',
 	PYTHON = 'text/x-python',
 	JAVA = 'text/x-java-source',
+	CPP_HDR = 'text/x-c++hdr',
 	CPP_SRC = 'text/x-c++src',
+	CSHARP = 'text/x-csharp',
+	HASKELL = 'text/x-haskell',
 	C_SRC = 'text/x-csrc',
 	C_HDR = 'text/x-chdr',
 	PHP = 'text/x-php',
@@ -182,6 +197,10 @@ export enum MimeTypeText {
 	DART = 'text/x-dart',
 	VUE = 'text/x-vue',
 	SVELTE = 'text/x-svelte',
-	LATEX = 'text/x-tex',
-	BIBTEX = 'text/x-bibtex'
+	TEX = 'text/x-tex',
+	TEX_APP = 'application/x-tex',
+	LATEX = 'application/x-latex',
+	BIBTEX = 'text/x-bibtex',
+	CUDA = 'text/x-cuda',
+	PROPERTIES = 'text/properties'
 }

tools/server/webui/src/lib/services/chat.ts

@@ -681,48 +681,6 @@ export class ChatService {
 	// Utilities
 	// ─────────────────────────────────────────────────────────────────────────────
 
-	/**
-	 * Get server properties - static method for API compatibility (to be refactored)
-	 */
-	static async getServerProps(): Promise<ApiLlamaCppServerProps> {
-		try {
-			const response = await fetch(`./props`, {
-				headers: getJsonHeaders()
-			});
-
-			if (!response.ok) {
-				throw new Error(`Failed to fetch server props: ${response.status}`);
-			}
-
-			const data = await response.json();
-			return data;
-		} catch (error) {
-			console.error('Error fetching server props:', error);
-			throw error;
-		}
-	}
-
-	/**
-	 * Get model information from /models endpoint (to be refactored)
-	 */
-	static async getModels(): Promise<ApiModelListResponse> {
-		try {
-			const response = await fetch(`./models`, {
-				headers: getJsonHeaders()
-			});
-
-			if (!response.ok) {
-				throw new Error(`Failed to fetch models: ${response.status} ${response.statusText}`);
-			}
-
-			const data = await response.json();
-			return data;
-		} catch (error) {
-			console.error('Error fetching models:', error);
-			throw error;
-		}
-	}
-
 	/**
 	 * Injects a system message at the beginning of the conversation if provided.
 	 * Checks for existing system messages to avoid duplication.

tools/server/webui/src/lib/services/models.ts

@@ -7,7 +7,7 @@ import { getJsonHeaders } from '$lib/utils';
  *
  * This service handles communication with model-related endpoints:
  * - `/v1/models` - OpenAI-compatible model list (MODEL + ROUTER mode)
- * - `/models` - Router-specific model management (ROUTER mode only)
+ * - `/models/load`, `/models/unload` - Router-specific model management (ROUTER mode only)
  *
  * **Responsibilities:**
  * - List available models
@@ -43,7 +43,7 @@ export class ModelsService {
 	 * Returns models with load status, paths, and other metadata
 	 */
 	static async listRouter(): Promise<ApiRouterModelsListResponse> {
-		const response = await fetch(`${base}/models`, {
+		const response = await fetch(`${base}/v1/models`, {
 			headers: getJsonHeaders()
 		});
 

tools/server/webui/src/lib/stores/conversations.svelte.ts

@@ -519,6 +519,19 @@ class ConversationsStore {
 		return await DatabaseService.getConversationMessages(convId);
 	}
 
+	/**
+	 * Imports conversations from provided data (without file picker)
+	 * @param data - Array of conversation data with messages
+	 * @returns Import result with counts
+	 */
+	async importConversationsData(
+		data: ExportedConversations
+	): Promise<{ imported: number; skipped: number }> {
+		const result = await DatabaseService.importConversations(data);
+		await this.loadConversations();
+		return result;
+	}
+
 	/**
 	 * Adds a message to the active messages array
 	 * Used by chatStore when creating new messages

tools/server/webui/src/lib/stores/settings.svelte.ts

@@ -370,6 +370,10 @@ class SettingsStore {
 		return { ...this.config };
 	}
 
+	canSyncParameter(key: string): boolean {
+		return ParameterSyncService.canSyncParameter(key);
+	}
+
 	/**
 	 * Get parameter information including source for a specific parameter
 	 */

tools/server/webui/src/lib/utils/file-type.ts

@@ -77,6 +77,13 @@ export function getFileTypeCategory(mimeType: string): FileTypeCategory | null {
 		case MimeTypeText.SVELTE:
 		case MimeTypeText.LATEX:
 		case MimeTypeText.BIBTEX:
+		case MimeTypeText.CUDA:
+		case MimeTypeText.CPP_HDR:
+		case MimeTypeText.CSHARP:
+		case MimeTypeText.HASKELL:
+		case MimeTypeText.PROPERTIES:
+		case MimeTypeText.TEX:
+		case MimeTypeText.TEX_APP:
 			return FileTypeCategory.TEXT;
 
 		default:
@@ -144,6 +151,12 @@ export function getFileTypeCategoryByExtension(filename: string): FileTypeCatego
 		case FileExtensionText.SVELTE:
 		case FileExtensionText.TEX:
 		case FileExtensionText.BIB:
+		case FileExtensionText.COMP:
+		case FileExtensionText.CU:
+		case FileExtensionText.CUH:
+		case FileExtensionText.HPP:
+		case FileExtensionText.HS:
+		case FileExtensionText.PROPERTIES:
 			return FileTypeCategory.TEXT;
 
 		default: