Merge branch 'upstream' into concedo_experimental

# Conflicts: # .devops/cann.Dockerfile # .devops/cpu.Dockerfile # .devops/cuda.Dockerfile # .devops/intel.Dockerfile # .devops/llama-cli-cann.Dockerfile # .devops/musa.Dockerfile # .devops/openvino.Dockerfile # .devops/rocm.Dockerfile # .devops/s390x.Dockerfile # .devops/vulkan.Dockerfile # .github/ISSUE_TEMPLATE/011-bug-results.yml # .github/ISSUE_TEMPLATE/019-bug-misc.yml # .github/workflows/build-and-test-snapdragon.yml # .github/workflows/docker.yml # .github/workflows/server-self-hosted.yml # .github/workflows/ui-ci.yml # .pi/gg/SYSTEM.md # README.md # common/arg.cpp # docs/backend/SYCL.md # docs/backend/snapdragon/CMakeUserPresets.json # docs/backend/snapdragon/README.md # docs/speculative.md # examples/save-load-state/save-load-state.cpp # ggml/src/ggml-hexagon/ggml-hexagon.cpp # ggml/src/ggml-hexagon/htp/CMakeLists.txt # ggml/src/ggml-hexagon/htp/htp-ctx.h # ggml/src/ggml-hexagon/htp/htp-ops.h # ggml/src/ggml-hexagon/htp/main.c # ggml/src/ggml-hexagon/htp/rope-ops.c # ggml/src/ggml-hexagon/htp/unary-ops.c # ggml/src/ggml-opencl/CMakeLists.txt # ggml/src/ggml-opencl/ggml-opencl.cpp # ggml/src/ggml-opencl/kernels/cvt.cl # ggml/src/ggml-sycl/ggml-sycl.cpp # ggml/src/ggml-webgpu/ggml-webgpu.cpp # ggml/src/ggml-webgpu/wgsl-shaders/gated_delta_net.wgsl # tools/cli/README.md # tools/server/README.md
2026-09-19 17:25:07 +02:00 · 2026-05-20 18:48:34 +08:00
parent 643862ac7d 585080d310
commit 7d987af23a
77 changed files with 1291 additions and 1048 deletions
@@ -30,7 +30,7 @@ int main(int argc, char ** argv) {
    if (!params.fit_params_print) {
        const common_params_fit_status status = common_fit_params(params.model.path.c_str(), &mparams, &cparams,
                params.tensor_split, params.tensor_buft_overrides.data(), params.fit_params_target.data(), params.fit_params_min_ctx,
-                params.verbosity >= 4 ? GGML_LOG_LEVEL_DEBUG : GGML_LOG_LEVEL_ERROR);
+                params.verbosity >= LOG_LEVEL_DEBUG ? GGML_LOG_LEVEL_DEBUG : GGML_LOG_LEVEL_ERROR);
        if (status != COMMON_PARAMS_FIT_STATUS_SUCCESS) {
            LOG_ERR("%s: failed to fit CLI arguments to free memory, exiting...\n", __func__);
            exit(1);
@@ -467,20 +467,26 @@ struct server_slot {
        const double n_gen_second = 1e3 / t_token_generation * n_decoded;

        SLT_INF(*this,
-                "\n"
-                "prompt eval time = %10.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n"
-                "       eval time = %10.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n"
+                "prompt eval time = %10.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
+                t_prompt_processing, n_prompt_tokens_processed, t_prompt, n_prompt_second);
+
+        SLT_INF(*this,
+                "       eval time = %10.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
+                t_token_generation, n_decoded, t_gen, n_gen_second);
+
+        SLT_INF(*this,
                "      total time = %10.2f ms / %5d tokens\n",
-                t_prompt_processing, n_prompt_tokens_processed, t_prompt, n_prompt_second,
-                t_token_generation, n_decoded, t_gen, n_gen_second,
                t_prompt_processing + t_token_generation, n_prompt_tokens_processed + n_decoded);

+        SLT_INF(*this,
+                "   graphs reused = %10d\n",
+                llama_perf_context(ctx_tgt).n_reused);
+
        if (n_draft_total > 0) {
            const float draft_ratio = (float) n_draft_accepted / n_draft_total;
-            SLT_CNT(*this,
-                    "draft acceptance rate = %0.5f (%5d accepted / %5d generated)\n",
-                    draft_ratio, n_draft_accepted, n_draft_total
-            );
+            SLT_INF(*this,
+                    "draft acceptance = %0.5f (%5d accepted / %5d generated)\n",
+                    draft_ratio, n_draft_accepted, n_draft_total);
        }

        common_speculative_print_stats(spec);
@@ -2583,9 +2589,9 @@ private:
                            llama_pos pos_next = slot.prompt.tokens.pos_next(n_past);

                            // the largest pos_min required for a checkpoint to be useful
-                            const auto pos_min_thold = std::max(0, pos_next - n_swa);
+                            const auto pos_min_thold = std::max(0, pos_next - n_swa - 1);

-                            if (n_past > 0 && n_past < slot.prompt.n_tokens()) {
+                            if (n_past > 0 && n_past <= slot.prompt.n_tokens()) {
                                const auto pos_min = llama_memory_seq_pos_min(llama_get_memory(ctx_tgt), slot.id);
                                if (pos_min == -1) {
                                    SLT_ERR(slot, "n_past = %d, slot.prompt.tokens.size() = %d, seq_id = %d, pos_min = %d\n", n_past, (int) slot.prompt.tokens.size(), slot.id, pos_min);
@@ -3885,6 +3891,7 @@ void server_routes::init_routes() {
            { "eos_token",                   meta->eos_token_str },
            { "build_info",                  meta->build_info },
            { "is_sleeping",                 queue_tasks.is_sleeping() },
+            { "cors_proxy_enabled",          params.ui_mcp_proxy || params.webui_mcp_proxy },
        };
        if (params.use_jinja) {
            if (!tmpl_tools.empty()) {
@@ -1165,6 +1165,7 @@ void server_models_routes::init_routes() {
                // Deprecated: use ui_settings instead (kept for backward compat)
                {"webui_settings",  webui_settings},
                {"build_info",     std::string(llama_build_info())},
+                {"cors_proxy_enabled", params.ui_mcp_proxy || params.webui_mcp_proxy},
            });
            return res;
        }
@@ -0,0 +1,2 @@
+VITE_PUBLIC_APP_NAME='llama-ui'
+# VITE_DEBUG='true'
@@ -25,4 +25,4 @@ vite.config.ts.timestamp-*

 *storybook.log
 storybook-static
-*.code-workspace
+*.code-workspace
@@ -20,9 +20,7 @@ export default ts.config(
 	prettier,
 	...svelte.configs.prettier,
 	{
-		languageOptions: {
-			globals: { ...globals.browser, ...globals.node }
-		},
+		languageOptions: { globals: { ...globals.browser, ...globals.node } },
 		rules: {
 			// typescript-eslint strongly recommend that you do not use the no-undef lint rule on TypeScript projects.
 			// see: https://typescript-eslint.io/troubleshooting/faqs/eslint/#i-get-errors-from-the-no-undef-rule-about-global-variables-not-being-defined-even-though-there-are-no-typescript-errors
@@ -30,6 +28,7 @@ export default ts.config(
 			'svelte/no-at-html-tags': 'off',
 			// This app uses hash-based routing (#/) where resolve() from $app/paths does not apply
 			'svelte/no-navigation-without-resolve': 'off',
+
 			// Enforce empty line at end of file
 			'eol-last': 'error'
 		}
@@ -2307,9 +2307,9 @@
 			}
 		},
 		"node_modules/@sveltejs/kit": {
-			"version": "2.59.1",
-			"resolved": "https://registry.npmjs.org/@sveltejs/kit/-/kit-2.59.1.tgz",
-			"integrity": "sha512-d8OON70AphLdDesuTIl//M2O6fRTIicX8aYv8vhCiYEhTTI2OboKqey0Hu1A4VFhqwgqtq0vKDmPFGkw8kKmgw==",
+			"version": "2.60.1",
+			"resolved": "https://registry.npmjs.org/@sveltejs/kit/-/kit-2.60.1.tgz",
+			"integrity": "sha512-mQjlkNo+rJvpln7V2IGY2j99BqhcFbS4UN0AQNKNYfhBAFZTuCDAdW3a1sgf330mvtNvsBXn3HpAhcmvdJTcIQ==",
 			"dev": true,
 			"license": "MIT",
 			"dependencies": {
@@ -2318,7 +2318,7 @@
 				"@types/cookie": "^0.6.0",
 				"acorn": "^8.14.1",
 				"cookie": "^0.6.0",
-				"devalue": "^5.6.4",
+				"devalue": "^5.8.1",
 				"esm-env": "^1.2.2",
 				"kleur": "^4.1.5",
 				"magic-string": "^0.30.5",
@@ -4296,9 +4296,9 @@
 			}
 		},
 		"node_modules/devalue": {
-			"version": "5.6.4",
-			"resolved": "https://registry.npmjs.org/devalue/-/devalue-5.6.4.tgz",
-			"integrity": "sha512-Gp6rDldRsFh/7XuouDbxMH3Mx8GMCcgzIb1pDTvNyn8pZGQ22u+Wa+lGV9dQCltFQ7uVw0MhRyb8XDskNFOReA==",
+			"version": "5.8.1",
+			"resolved": "https://registry.npmjs.org/devalue/-/devalue-5.8.1.tgz",
+			"integrity": "sha512-4CXDYRBGqN+57wVJkuXBYmpAVUSg3L6JAQa/DFqm238G73E1wuyc/JhGQJzN7vUf/CMphYau2zXbfWzDR5aTEw==",
 			"license": "MIT"
 		},
 		"node_modules/devlop": {
@@ -4856,12 +4856,12 @@
 			}
 		},
 		"node_modules/express-rate-limit": {
-			"version": "8.5.0",
-			"resolved": "https://registry.npmjs.org/express-rate-limit/-/express-rate-limit-8.5.0.tgz",
-			"integrity": "sha512-XKhFohWaSBdVJNTi5TaHziqnPkv04I9UQV6q1Wy7Ui6GGQZVW12ojDFwqer14EvCXxjvPG0CyWXx7cAXpALB4Q==",
+			"version": "8.5.2",
+			"resolved": "https://registry.npmjs.org/express-rate-limit/-/express-rate-limit-8.5.2.tgz",
+			"integrity": "sha512-5Kb34ipNX694DH48vN9irak1Qx30nb0PLYHXfJgw4YEjiC3ZEmZJhwOp+VfiCYwFzvFTdB9QkArYS5kXa2cx2A==",
 			"license": "MIT",
 			"dependencies": {
-				"ip-address": "10.1.0"
+				"ip-address": "^10.2.0"
 			},
 			"engines": {
 				"node": ">= 16"
@@ -4909,9 +4909,9 @@
 			"license": "MIT"
 		},
 		"node_modules/fast-uri": {
-			"version": "3.1.0",
-			"resolved": "https://registry.npmjs.org/fast-uri/-/fast-uri-3.1.0.tgz",
-			"integrity": "sha512-iPeeDKJSWf4IEOasVVrknXpaBV0IApz/gp7S2bb7Z4Lljbl2MGJRqInZiUrQwV16cpzw/D3S5j5Julj/gT52AA==",
+			"version": "3.1.2",
+			"resolved": "https://registry.npmjs.org/fast-uri/-/fast-uri-3.1.2.tgz",
+			"integrity": "sha512-rVjf7ArG3LTk+FS6Yw81V1DLuZl1bRbNrev6Tmd/9RaroeeRRJhAt7jg/6YFxbvAQXUCavSoZhPPj6oOx+5KjQ==",
 			"funding": [
 				{
 					"type": "github",
@@ -5541,9 +5541,9 @@
 			}
 		},
 		"node_modules/hono": {
-			"version": "4.12.14",
-			"resolved": "https://registry.npmjs.org/hono/-/hono-4.12.14.tgz",
-			"integrity": "sha512-am5zfg3yu6sqn5yjKBNqhnTX7Cv+m00ox+7jbaKkrLMRJ4rAdldd1xPd/JzbBWspqaQv6RSTrgFN95EsfhC+7w==",
+			"version": "4.12.19",
+			"resolved": "https://registry.npmjs.org/hono/-/hono-4.12.19.tgz",
+			"integrity": "sha512-xa3eYXYXx68XTT4hZ7dRzsXBhaq85ToSrlUJNoR0gwz/1Ap/CNwX47wfvV7pc/xWhjKVVkLT7zBJy8chhNguqQ==",
 			"license": "MIT",
 			"engines": {
 				"node": ">=16.9.0"
@@ -5722,9 +5722,9 @@
 			"license": "MIT"
 		},
 		"node_modules/ip-address": {
-			"version": "10.1.0",
-			"resolved": "https://registry.npmjs.org/ip-address/-/ip-address-10.1.0.tgz",
-			"integrity": "sha512-XXADHxXmvT9+CRxhXg56LJovE+bmWnEWB78LB83VZTprKTmaC5QfruXocxzTZ2Kl0DNwKuBdlIhjL8LeY8Sf8Q==",
+			"version": "10.2.0",
+			"resolved": "https://registry.npmjs.org/ip-address/-/ip-address-10.2.0.tgz",
+			"integrity": "sha512-/+S6j4E9AHvW9SWMSEY9Xfy66O5PWvVEJ08O0y5JGyEKQpojb0K0GKpz/v5HJ/G0vi3D2sjGK78119oXZeE0qA==",
 			"license": "MIT",
 			"engines": {
 				"node": ">= 12"
@@ -6008,9 +6008,9 @@
 			}
 		},
 		"node_modules/katex": {
-			"version": "0.16.22",
-			"resolved": "https://registry.npmjs.org/katex/-/katex-0.16.22.tgz",
-			"integrity": "sha512-XCHRdUw4lf3SKBaJe4EvgqIuWwkPSo9XoeO8GjQW94Bp7TWv9hNhzZjZ+OH9yf1UmLygb7DIT5GSFQiyt16zYg==",
+			"version": "0.16.47",
+			"resolved": "https://registry.npmjs.org/katex/-/katex-0.16.47.tgz",
+			"integrity": "sha512-Eeo8Ys1doU1z+x8AZsPpQu+p/QcZBI5PeOo7QGQdy2x2m0MU/hYagBbGOmXwr5KVbEfVuWv9LpnQWeehogurjg==",
 			"dev": true,
 			"funding": [
 				"https://opencollective.com/katex",
@@ -9245,9 +9245,9 @@
 			}
 		},
 		"node_modules/svelte": {
-			"version": "5.55.1",
-			"resolved": "https://registry.npmjs.org/svelte/-/svelte-5.55.1.tgz",
-			"integrity": "sha512-QjvU7EFemf6mRzdMGlAFttMWtAAVXrax61SZYHdkD6yoVGQ89VeyKfZD4H1JrV1WLmJBxWhFch9H6ig/87VGjw==",
+			"version": "5.55.7",
+			"resolved": "https://registry.npmjs.org/svelte/-/svelte-5.55.7.tgz",
+			"integrity": "sha512-ymI5ykLPwIHW839E053FQbI1G+jnRFJEw3Kv5Y4njixVWywQBx+NUFpkkKyk5LIb36Fg9DVXSYpqiGekLD0hyw==",
 			"license": "MIT",
 			"dependencies": {
 				"@jridgewell/remapping": "^2.3.4",
@@ -9259,7 +9259,7 @@
 				"aria-query": "5.3.1",
 				"axobject-query": "^4.1.0",
 				"clsx": "^2.1.1",
-				"devalue": "^5.6.4",
+				"devalue": "^5.8.1",
 				"esm-env": "^1.2.1",
 				"esrap": "^2.2.4",
 				"is-reference": "^3.0.3",
@@ -10606,9 +10606,9 @@
 			"license": "ISC"
 		},
 		"node_modules/ws": {
-			"version": "8.18.3",
-			"resolved": "https://registry.npmjs.org/ws/-/ws-8.18.3.tgz",
-			"integrity": "sha512-PEIGCY5tSlUt50cqyMXfCzX+oOPqN0vuGqWzbcJ2xvnkzkq46oOpz7dQaTDBdfICb4N14+GARUDw2XV2N4tvzg==",
+			"version": "8.20.1",
+			"resolved": "https://registry.npmjs.org/ws/-/ws-8.20.1.tgz",
+			"integrity": "sha512-It4dO0K5v//JtTXuPkfEOaI3uUN87iYPnqo/ZzqCoG3g8uhA66QUMs/SrM0YK7/NAu+r4LMh/9dq2A7k+rHs+w==",
 			"dev": true,
 			"license": "MIT",
 			"engines": {
@@ -1,6 +1,7 @@
@import 'tailwindcss';
-@source ".";
-
+@source '.';
+@plugin '@tailwindcss/forms';
+@plugin '@tailwindcss/typography';
@import 'tw-animate-css';

@custom-variant dark (&:is(.dark *));
@@ -15,6 +15,7 @@

 		{#if videoSrc}
 			<video controls class="mb-4 w-full" src={videoSrc}>
+				<track kind="captions" src="" />
 				Your browser does not support the video element.
 			</video>
 		{:else}
@@ -7,7 +7,6 @@
 	import { activeMessages } from '$lib/stores/conversations.svelte';

 	interface Props {
-		currentModel?: string;
 		disabled?: boolean;
 		forceForegroundText?: boolean;
 		hasAudioModality?: boolean;
@@ -20,7 +19,6 @@
 	}

 	let {
-		currentModel,
 		disabled = false,
 		forceForegroundText = false,
 		hasAudioModality = $bindable(false),
@@ -41,14 +39,28 @@

 	let lastSyncedConversationModel: string | null = null;

+	let selectorModel = $derived(conversationModel ?? modelsStore.selectedModelName ?? null);
+
 	$effect(() => {
 		if (conversationModel && conversationModel !== lastSyncedConversationModel) {
-			lastSyncedConversationModel = conversationModel;
+			if (modelOptions().some((m) => m.model === conversationModel)) {
+				modelsStore.selectedModelName = conversationModel;
+				modelsStore.selectModelByName(conversationModel);
+			} else {
+				modelsStore.selectedModelName = null;
+				modelsStore.clearSelection();
+			}

-			modelsStore.selectModelByName(conversationModel);
-		} else if (isRouter && !modelsStore.selectedModelId && modelsStore.loadedModelIds.length > 0) {
+			lastSyncedConversationModel = conversationModel;
+		} else if (
+			isRouter &&
+			!modelsStore.selectedModelId &&
+			modelsStore.loadedModelIds.length > 0 &&
+			activeMessages().length > 0 &&
+			!conversationModel
+		) {
 			lastSyncedConversationModel = null;
-			// auto-select the first loaded model only when nothing is selected yet
+
 			const first = modelOptions().find((m) => modelsStore.loadedModelIds.includes(m.model));

 			if (first) modelsStore.selectModelById(first.id);
@@ -151,7 +163,7 @@
 	<ModelsSelectorSheet
 		disabled={disabled || isOffline}
 		bind:this={selectorModelRef}
-		{currentModel}
+		currentModel={selectorModel}
 		{forceForegroundText}
 		{useGlobalSelection}
 	/>
@@ -159,7 +171,7 @@
 	<ModelsSelectorDropdown
 		disabled={disabled || isOffline}
 		bind:this={selectorModelRef}
-		{currentModel}
+		currentModel={selectorModel}
 		{forceForegroundText}
 		{useGlobalSelection}
 	/>
@@ -162,7 +162,7 @@
 			return;
 		}

-		if (import.meta.env.DEV) {
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
 			console.log('[ChatFormPickerMcpPrompts] Fetching completions for:', {
 				serverName: selectedPrompt.serverName,
 				promptName: selectedPrompt.name,
@@ -181,7 +181,7 @@
 				value
 			);

-			if (import.meta.env.DEV) {
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
 				console.log('[ChatFormPickerMcpPrompts] Autocomplete result:', {
 					argName,
 					value,
@@ -1,20 +1,20 @@
 <script lang="ts">
-	import { Trash2, AlertTriangle, RefreshCw } from '@lucide/svelte';
+	import { Trash2 } from '@lucide/svelte';
 	import { afterNavigate } from '$app/navigation';
 	import { page } from '$app/state';
-	import { fadeInView } from '$lib/actions/fade-in-view.svelte';
 	import {
 		ChatScreenForm,
 		ChatMessages,
 		ChatScreenDragOverlay,
 		ChatScreenProcessingInfo,
+		ChatScreenActionScrollDown,
 		DialogEmptyFileAlert,
 		DialogFileUploadError,
 		DialogChatError,
 		ServerLoadingSplash,
-		DialogConfirmation
+		DialogConfirmation,
+		ChatScreenServerError
 	} from '$lib/components/app';
-	import * as Alert from '$lib/components/ui/alert';
 	import { setProcessingInfoContext } from '$lib/contexts';
 	import { ErrorDialogType } from '$lib/enums';
 	import { createAutoScrollController } from '$lib/hooks/use-auto-scroll.svelte';
@@ -34,11 +34,12 @@
 		activeConversation
 	} from '$lib/stores/conversations.svelte';
 	import { config } from '$lib/stores/settings.svelte';
-	import { serverLoading, serverError, serverStore, isRouterMode } from '$lib/stores/server.svelte';
+	import { serverLoading, serverError, isRouterMode } from '$lib/stores/server.svelte';
 	import { modelsStore, modelOptions, selectedModelId } from '$lib/stores/models.svelte';
 	import { isFileTypeSupported, filterFilesByModalities } from '$lib/utils';
 	import { parseFilesToMessageExtras, processFilesToChatUploaded } from '$lib/utils/browser-only';
 	import { onMount } from 'svelte';
+	import ChatScreenGreeting from './ChatScreenGreeting.svelte';

 	let { showCenteredEmpty = false } = $props();

@@ -67,6 +68,8 @@

 	let showEmptyFileDialog = $state(false);

+	let processingInfoVisible = $state(false);
+
 	let emptyFileNames = $state<string[]>([]);

 	let initialMessage = $state('');
@@ -174,6 +177,10 @@
 		showDeleteDialog = false;
 	}

+	function handleProcessingInfoVisibility(visible: boolean) {
+		processingInfoVisible = visible;
+	}
+
 	function handleDragEnter(event: DragEvent) {
 		event.preventDefault();

@@ -338,7 +345,9 @@
 	});

 	function handleMessagesReady() {
-		if (!disableAutoScroll && !autoScroll.userScrolledUp) {
+		if (disableAutoScroll) return;
+
+		if (!autoScroll.userScrolledUp) {
 			requestAnimationFrame(() => {
 				autoScroll.scrollToBottom('instant');
 			});
@@ -392,59 +401,32 @@
 			{#if !isEmpty}
 				<ChatMessages
 					messages={activeMessages()}
+					onMessagesReady={handleMessagesReady}
 					onUserAction={() => {
 						autoScroll.enable();
 						if (!autoScroll.userScrolledUp) {
 							autoScroll.scrollToBottom();
 						}
 					}}
-					onMessagesReady={handleMessagesReady}
 				/>
 			{/if}

 			<div
-				class="pointer-events-none {isEmpty
-					? 'absolute bottom-[calc(50dvh-7rem)]'
-					: 'sticky bottom-4'} right-4 left-4 mt-auto pt-16 transition-all duration-200"
+				class={[
+					'pointer-events-none sticky right-4 left-4 mt-auto transition-all duration-200',
+					isEmpty ? 'bottom-[calc(50dvh-7rem)]' : 'bottom-4 pt-24 md:pt-32'
+				]}
 			>
-				{#if isEmpty}
-					<div class="mb-8 px-4 text-center" use:fadeInView={{ duration: 300 }}>
-						<h1 class="mb-2 text-2xl font-semibold tracking-tight md:text-3xl">Hello there</h1>
+				<ChatScreenGreeting {isEmpty} />

-						<p class="text-muted-foreground md:text-lg">
-							{serverStore.props?.modalities?.audio
-								? 'Record audio, type a message '
-								: 'Type a message'} or upload files to get started
-						</p>
-					</div>
-				{/if}
+				<ChatScreenActionScrollDown
+					container={chatScrollContainer}
+					hasProcessingInfoVisible={processingInfoVisible}
+				/>

-				{#if page.params.id}
-					<ChatScreenProcessingInfo />
-				{/if}
+				<ChatScreenProcessingInfo onVisibilityChange={handleProcessingInfoVisibility} />

-				{#if hasPropsError}
-					<div
-						class="pointer-events-auto mx-auto mb-4 max-w-[48rem] px-1"
-						use:fadeInView={{ y: 10, duration: 250 }}
-					>
-						<Alert.Root variant="destructive">
-							<AlertTriangle class="h-4 w-4" />
-							<Alert.Title class="flex items-center justify-between">
-								<span>Server unavailable</span>
-								<button
-									onclick={() => serverStore.fetch()}
-									disabled={isServerLoading}
-									class="flex items-center gap-1.5 rounded-lg bg-destructive/20 px-2 py-1 text-xs font-medium hover:bg-destructive/30 disabled:opacity-50"
-								>
-									<RefreshCw class="h-3 w-3 {isServerLoading ? 'animate-spin' : ''}" />
-									{isServerLoading ? 'Retrying...' : 'Retry'}
-								</button>
-							</Alert.Title>
-							<Alert.Description>{serverError()}</Alert.Description>
-						</Alert.Root>
-					</div>
-				{/if}
+				<ChatScreenServerError />

 				<div class="conversation-chat-form pointer-events-auto rounded-t-3xl">
 					<ChatScreenForm
@@ -0,0 +1,61 @@
+<script lang="ts">
+	import { ArrowDown } from '@lucide/svelte';
+	import { Button } from '$lib/components/ui/button';
+
+	interface Props {
+		container: HTMLDivElement | undefined;
+		hasProcessingInfoVisible: boolean;
+	}
+
+	let { container, hasProcessingInfoVisible }: Props = $props();
+
+	let show = $state(false);
+
+	let buttonBottom = $derived(hasProcessingInfoVisible ? '2rem' : '0');
+
+	function checkVisibility() {
+		if (!container) return;
+		const { scrollTop, scrollHeight, clientHeight } = container;
+		const distanceFromBottom = scrollHeight - clientHeight - scrollTop;
+		show = distanceFromBottom > clientHeight * 0.5;
+	}
+
+	function scrollToBottom() {
+		if (container) {
+			container.scrollTo({
+				top: container.scrollHeight,
+				behavior: 'smooth'
+			});
+		}
+	}
+
+	$effect(() => {
+		const c = container;
+		if (c) {
+			c.addEventListener('scroll', checkVisibility);
+			checkVisibility();
+			return () => {
+				c.removeEventListener('scroll', checkVisibility);
+			};
+		}
+	});
+</script>
+
+<div
+	class="pointer-events-{show
+		? 'auto'
+		: 'none'} relative z-50 mx-auto mb-4 flex max-w-[48rem] justify-center"
+>
+	<Button
+		onclick={scrollToBottom}
+		variant="secondary"
+		size="icon"
+		class="pointer-events-all absolute h-10 w-10 rounded-full bg-background/80 shadow-lg backdrop-blur-sm transition-all duration-200 hover:bg-muted/80"
+		style="bottom: {buttonBottom}; transform: translateY({show ? '0' : '2rem'}); opacity: {show
+			? 1
+			: 0};"
+		aria-label="Scroll to bottom"
+	>
+		<ArrowDown class="h-4 w-4" />
+	</Button>
+</div>
@@ -0,0 +1,25 @@
+<script lang="ts">
+	import { fadeInView } from '$lib/actions/fade-in-view.svelte';
+	import { serverStore } from '$lib/stores/server.svelte';
+
+	interface Props {
+		isEmpty: boolean;
+	}
+
+	let { isEmpty = false }: Props = $props();
+</script>
+
+<div
+	class={[
+		'pointer-events-none mb-4 hidden px-4 text-center',
+		isEmpty && 'pointer-events-auto block!'
+	]}
+	use:fadeInView={{ duration: 300 }}
+>
+	<h1 class="mb-2 text-2xl font-semibold tracking-tight md:text-3xl">Hello there</h1>
+
+	<p class="text-muted-foreground md:text-lg">
+		{serverStore.props?.modalities?.audio ? 'Record audio, type a message ' : 'Type a message'} or upload
+		files to get started
+	</p>
+</div>
@@ -6,6 +6,7 @@
 	import { activeMessages, activeConversation } from '$lib/stores/conversations.svelte';
 	import { config } from '$lib/stores/settings.svelte';
 	import { getProcessingInfoContext } from '$lib/contexts';
+	import { page } from '$app/state';

 	const processingState = useProcessingState();
 	const processingInfoCtx = getProcessingInfoContext();
@@ -16,6 +17,14 @@
 	let isStreaming = $derived(isChatStreaming());
 	let processingDetails = $derived(processingState.getTechnicalDetails());

+	let processingVisible = $derived(processingDetails.length > 0);
+
+	let { onVisibilityChange }: { onVisibilityChange?: (visible: boolean) => void } = $props();
+
+	$effect(() => {
+		onVisibilityChange?.(processingVisible);
+	});
+
 	$effect(() => {
 		const conversation = activeConversation();

@@ -60,9 +69,12 @@
 </script>

 <div
-	class={['chat-processing-info-container pointer-events-none', showProcessingInfo && 'visible']}
+	class={[
+		'chat-processing-info-container pointer-events-none relative',
+		page.params.id && showProcessingInfo && 'visible'
+	]}
 >
-	<div class="chat-processing-info-content">
+	<div class="chat-processing-info-content absolute bottom-4 left-1/2 -translate-x-1/2">
 		{#each processingDetails as detail (detail)}
 			<span class="chat-processing-info-detail pointer-events-auto backdrop-blur-sm">{detail}</span>
 		{/each}
@@ -0,0 +1,34 @@
+<script lang="ts">
+	import { AlertTriangle, RefreshCw } from '@lucide/svelte';
+	import { fadeInView } from '$lib/actions/fade-in-view.svelte';
+	import * as Alert from '$lib/components/ui/alert';
+	import { serverError, serverLoading, serverStore } from '$lib/stores/server.svelte';
+
+	let hasError = $derived(!!serverError());
+</script>
+
+{#if hasError}
+	<div
+		class="pointer-events-auto mx-auto mb-4 max-w-[48rem] px-1"
+		use:fadeInView={{ y: 10, duration: 250 }}
+	>
+		<Alert.Root variant="destructive">
+			<AlertTriangle class="h-4 w-4" />
+
+			<Alert.Title class="flex items-center justify-between">
+				<span>Server unavailable</span>
+
+				<button
+					onclick={() => serverStore.fetch()}
+					disabled={serverLoading()}
+					class="flex items-center gap-1.5 rounded-lg bg-destructive/20 px-2 py-1 text-xs font-medium hover:bg-destructive/30 disabled:opacity-50"
+				>
+					<RefreshCw class="h-3 w-3 {serverLoading() ? 'animate-spin' : ''}" />
+					{serverLoading() ? 'Retrying...' : 'Retry'}
+				</button>
+			</Alert.Title>
+
+			<Alert.Description>{serverError()}</Alert.Description>
+		</Alert.Root>
+	</div>
+{/if}
@@ -667,3 +667,17 @@ export { default as ChatScreenForm } from './ChatScreen/ChatScreenForm.svelte';
 * Only visible when `isCurrentConversationLoading` is true.
 */
 export { default as ChatScreenProcessingInfo } from './ChatScreen/ChatScreenProcessingInfo.svelte';
+
+/**
+ * Scroll-to-bottom action button. Displays a floating button when the user
+ * has scrolled up more than half a viewport height from the bottom.
+ * Takes the chat container element as a prop to manage scroll state internally.
+ */
+export { default as ChatScreenActionScrollDown } from './ChatScreen/ChatScreenActionScrollDown.svelte';
+
+/**
+ * Server error alert displayed when the server is unreachable.
+ * Shows the error message with a retry button.
+ * Rendered inside ChatScreen when `serverError` store has a value.
+ */
+export { default as ChatScreenServerError } from './ChatScreen/ChatScreenServerError.svelte';
@@ -28,7 +28,7 @@
 		SETTINGS_KEYS
 	} from '$lib/constants';
 	import { ColorMode, UrlProtocol } from '$lib/enums';
-	import { FileTypeText } from '$lib/enums/files';
+	import { FileTypeText } from '$lib/enums/files.enums';
 	import { highlightCode, detectIncompleteCodeBlock, type IncompleteCodeBlock } from '$lib/utils';
 	import '$styles/katex-custom.scss';
 	import githubDarkCss from 'highlight.js/styles/github-dark.css?inline';
@@ -17,7 +17,7 @@
 	} from '$lib/constants';
 	import { RouterService } from '$lib/services/router.service';
 	import { setMode } from 'mode-watcher';
-	import { ColorMode } from '$lib/enums/ui';
+	import { ColorMode } from '$lib/enums/ui.enums';
 	import { fade } from 'svelte/transition';
 	import { goto } from '$app/navigation';
 	import { page } from '$app/state';
@@ -6,7 +6,7 @@
 	import * as Select from '$lib/components/ui/select';
 	import { Textarea } from '$lib/components/ui/textarea';
 	import { SETTING_CONFIG_INFO, SETTINGS_KEYS } from '$lib/constants';
-	import { SettingsFieldType } from '$lib/enums/settings';
+	import { SettingsFieldType } from '$lib/enums/settings.enums';
 	import { settingsStore } from '$lib/stores/settings.svelte';
 	import { serverStore } from '$lib/stores/server.svelte';
 	import { modelsStore, selectedModelName, propsCacheVersion } from '$lib/stores/models.svelte';
@@ -2,7 +2,7 @@ import { Zap, Globe, Radio } from '@lucide/svelte';
 import { MCPTransportType } from '$lib/enums';
 import type { ClientCapabilities, Implementation } from '$lib/types';
 import type { Component } from 'svelte';
-import { MimeTypeImage } from '$lib/enums/files';
+import { MimeTypeImage } from '$lib/enums/files.enums';

 export const DEFAULT_CLIENT_VERSION = '1.0.0';
 export const MCP_CLIENT_NAME = 'llama-ui-mcp';
@@ -1,5 +1,5 @@
-import { ColorMode } from '$lib/enums/ui';
-import { SettingsFieldType } from '$lib/enums/settings';
+import { ColorMode } from '$lib/enums/ui.enums';
+import { SettingsFieldType } from '$lib/enums/settings.enums';
 import { SyncableParameterType } from '$lib/enums';
 import {
 	Funnel,
@@ -18,7 +18,7 @@ import {
 	MimeTypeApplication,
 	MimeTypeText
 } from '$lib/enums';
-import { FileExtensionVideo, FileTypeVideo } from '$lib/enums/files';
+import { FileExtensionVideo, FileTypeVideo } from '$lib/enums/files.enums';

 // File type configuration using enums
 export const AUDIO_FILE_TYPES = {
@@ -1,4 +1,4 @@
-import { ToolSource } from '$lib/enums/tools';
+import { ToolSource } from '$lib/enums/tools.enums';

 export const TOOL_GROUP_LABELS = {
 	[ToolSource.BUILTIN]: 'Built-in',
@@ -4,9 +4,9 @@ export {
 	AttachmentItemEnabledWhen,
 	AttachmentAction,
 	AttachmentItemVisibleWhen
-} from './attachment';
+} from './attachment.enums';

-export { AgenticSectionType, ToolCallType } from './agentic';
+export { AgenticSectionType, ToolCallType } from './agentic.enums';

 export {
 	ChatMessageStatsView,
@@ -17,7 +17,7 @@ export {
 	MessageType,
 	PdfViewMode,
 	ReasoningFormat
-} from './chat';
+} from './chat.enums';

 export {
 	FileTypeCategory,
@@ -38,7 +38,7 @@ export {
 	MimeTypeImage,
 	MimeTypeText,
 	SpecialFileType
-} from './files';
+} from './files.enums';

 export {
 	MCPConnectionPhase,
@@ -48,16 +48,16 @@ export {
 	MCPContentType,
 	MCPRefType,
 	JsonSchemaType
-} from './mcp';
+} from './mcp.enums';

-export { ModelModality } from './model';
+export { ModelModality } from './model.enums';

-export { ServerRole, ServerModelStatus } from './server';
+export { ServerRole, ServerModelStatus } from './server.enums';

-export { ParameterSource, SyncableParameterType, SettingsFieldType } from './settings';
+export { ParameterSource, SyncableParameterType, SettingsFieldType } from './settings.enums';

-export { ColorMode, HtmlInputType, McpPromptVariant, TooltipSide, UrlProtocol } from './ui';
+export { ColorMode, HtmlInputType, McpPromptVariant, TooltipSide, UrlProtocol } from './ui.enums';

-export { KeyboardKey } from './keyboard';
+export { KeyboardKey } from './keyboard.enums';

-export { ToolSource, ToolPermissionDecision, ToolResponseField } from './tools';
+export { ToolSource, ToolPermissionDecision, ToolResponseField } from './tools.enums';
@@ -100,6 +100,14 @@ export class AutoScrollController {
 		this._autoScrollEnabled = true;
 	}

+	/**
+	 * Resets scroll state when switching conversations.
+	 */
+	resetScrollState(): void {
+		this._userScrolledUp = false;
+		this._autoScrollEnabled = true;
+	}
+
 	/**
 	 * Starts the auto-scroll interval for continuous scrolling during streaming.
 	 */
@@ -66,7 +66,6 @@ export function useModelsSelector(opts: UseModelsSelectorOptions): UseModelsSele
 	const serverModel = $derived(singleModelName());

 	const currentModel = $derived(opts.currentModel());
-	const useGlobalSelection = $derived(opts.useGlobalSelection?.() ?? false);
 	const onModelChange = $derived(opts.onModelChange?.());

 	const isHighlightedCurrentModelActive = $derived.by(() => {
@@ -128,6 +127,7 @@ export function useModelsSelector(opts: UseModelsSelectorOptions): UseModelsSele

 		if (onModelChange) {
 			const result = await onModelChange(option.id, option.model);
+
 			if (result === false) {
 				shouldCloseMenu = false;
 			}
@@ -142,12 +142,14 @@ export function useModelsSelector(opts: UseModelsSelectorOptions): UseModelsSele
 				const textarea = document.querySelector<HTMLTextAreaElement>(
 					'[data-slot="chat-form"] textarea'
 				);
+
 				textarea?.focus();
 			});
 		}

 		if (!onModelChange && isRouter && !modelsStore.isModelLoaded(option.model)) {
 			isLoadingModel = true;
+
 			modelsStore
 				.loadModel(option.model)
 				.catch((error) => console.error('Failed to load model:', error))
@@ -158,6 +160,7 @@ export function useModelsSelector(opts: UseModelsSelectorOptions): UseModelsSele
 	function getDisplayOption(): ModelOption | undefined {
 		if (!isRouter) {
 			const displayModel = serverModel || currentModel;
+
 			if (displayModel) {
 				return {
 					id: serverModel ? 'current' : 'offline-current',
@@ -166,12 +169,8 @@ export function useModelsSelector(opts: UseModelsSelectorOptions): UseModelsSele
 					capabilities: []
 				};
 			}
-			return undefined;
-		}

-		if (useGlobalSelection && activeId) {
-			const selected = options.find((option) => option.id === activeId);
-			if (selected) return selected;
+			return undefined;
 		}

 		if (currentModel) {
@@ -183,6 +182,7 @@ export function useModelsSelector(opts: UseModelsSelectorOptions): UseModelsSele
 					capabilities: []
 				};
 			}
+
 			return options.find((option) => option.model === currentModel);
 		}

@@ -197,57 +197,77 @@ export function useModelsSelector(opts: UseModelsSelectorOptions): UseModelsSele
 		get options() {
 			return options;
 		},
+
 		get loading() {
 			return loading;
 		},
+
 		get updating() {
 			return updating;
 		},
+
 		get activeId() {
 			return activeId;
 		},
+
 		get isRouter() {
 			return isRouter;
 		},
+
 		get serverModel() {
 			return serverModel;
 		},
+
 		get isHighlightedCurrentModelActive() {
 			return isHighlightedCurrentModelActive;
 		},
+
 		get isCurrentModelInCache() {
 			return isCurrentModelInCache;
 		},
+
 		get filteredOptions() {
 			return filteredOptions;
 		},
+
 		get groupedFilteredOptions() {
 			return groupedFilteredOptions;
 		},
+
 		get isLoadingModel() {
 			return isLoadingModel;
 		},
+
 		get searchTerm() {
 			return searchTerm;
 		},
+
 		get showModelDialog() {
 			return showModelDialog;
 		},
+
 		get infoModelId() {
 			return infoModelId;
 		},
+
 		setSearchTerm(value: string) {
 			searchTerm = value;
 		},
+
 		setShowModelDialog(value: boolean) {
 			showModelDialog = value;
 		},
+
 		handleInfoClick,
+
 		handleSelect,
+
 		handleOpenChange,
+
 		isFavorite(model: string) {
 			return modelsStore.favoriteModelIds.has(model);
 		},
+
 		getDisplayOption
 	};
 }
@@ -392,7 +392,7 @@ export class MCPService {

 			const url = new URL(config.url);

-			if (import.meta.env.DEV) {
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
 				console.log(`[MCPService] Creating WebSocket transport for ${url.href}`);
 			}

@@ -413,12 +413,12 @@ export class MCPService {
 			onLog
 		);

-		if (useProxy && import.meta.env.DEV) {
+		if (useProxy && import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
 			console.log(`[MCPService] Using CORS proxy for ${config.url} -> ${url.href}`);
 		}

 		try {
-			if (import.meta.env.DEV) {
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
 				console.log(`[MCPService] Creating StreamableHTTP transport for ${url.href}`);
 			}

@@ -520,7 +520,7 @@ export class MCPService {
 			)
 		);

-		if (import.meta.env.DEV) {
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
 			console.log(`[MCPService][${serverName}] Creating transport...`);
 		}

@@ -560,6 +560,22 @@ export class MCPService {
 		);

 		const runtimeErrorHandler = (error: Error) => {
+			// Ignore errors that are expected when the SDK's transport is closed,
+			// or when connecting to servers that don't support SSE (stateless-only
+			// endpoints returning 405). The SDK wraps the original AbortError in
+			// a new Error with the message "SSE stream disconnected: AbortError",
+			// and also produces "Cannot cancel a stream locked by a reader".
+			// DOMException is thrown by the browser when aborting fetch requests.
+			const msg = error.message || String(error);
+			if (
+				error.name === 'AbortError' ||
+				error instanceof DOMException ||
+				msg.includes('SSE stream disconnected') ||
+				msg.includes('stream locked by a reader') ||
+				msg.includes('The operation was aborted')
+			) {
+				return;
+			}
 			console.error(`[MCPService][${serverName}] Protocol error after initialize:`, error);
 		};

@@ -658,7 +674,10 @@ export class MCPService {
 			this.createLog(MCPConnectionPhase.LISTING_TOOLS, 'Listing available tools...')
 		);

-		console.log(`[MCPService][${serverName}] Connected, listing tools...`);
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
+			console.log(`[MCPService][${serverName}] Connected, listing tools...`);
+		}
+
 		const tools = await this.listTools({
 			client,
 			transport,
@@ -680,10 +699,11 @@ export class MCPService {
 				`Connection established with ${tools.length} tools (${connectionTimeMs}ms)`
 			)
 		);
-
-		console.log(
-			`[MCPService][${serverName}] Initialization complete with ${tools.length} tools in ${connectionTimeMs}ms`
-		);
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
+			console.log(
+				`[MCPService][${serverName}] Initialization complete with ${tools.length} tools in ${connectionTimeMs}ms`
+			);
+		}

 		return {
 			client,
@@ -709,9 +729,22 @@ export class MCPService {
 	 * @param connection - The active MCP connection to close
 	 */
 	static async disconnect(connection: MCPConnection): Promise<void> {
-		console.log(`[MCPService][${connection.serverName}] Disconnecting...`);
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
+			console.log(`[MCPService][${connection.serverName}] Disconnecting...`);
+		}
+
 		try {
-			// Prevent reconnection on voluntary disconnect
+			// Terminate the session first for streamable-http transports to cleanly
+			// close streams, matching the inspector's disconnect flow.
+			if (connection.transport instanceof StreamableHTTPClientTransport) {
+				await connection.transport.terminateSession();
+			}
+
+			// Clear error handlers before closing to prevent noise from expected
+			// abort errors during shutdown. The inspector avoids this entirely
+			// by not setting onerror, but since we use it for protocol logging,
+			// we must clear it before disconnect.
+			connection.client.onerror = undefined;
 			if (connection.transport.onclose) {
 				connection.transport.onclose = undefined;
 			}
@@ -1078,7 +1111,9 @@ export class MCPService {
 		try {
 			await connection.client.unsubscribeResource({ uri });

-			console.log(`[MCPService][${connection.serverName}] Unsubscribed from resource: ${uri}`);
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
+				console.log(`[MCPService][${connection.serverName}] Unsubscribed from resource: ${uri}`);
+			}
 		} catch (error) {
 			console.error(
 				`[MCPService][${connection.serverName}] Failed to unsubscribe from resource:`,
@@ -119,7 +119,8 @@ const localStorageMigration: Migration = {
 			// Only migrate if new key doesn't already exist
 			const newValue = localStorage.getItem(newKey);
 			if (newValue !== null) {
-				console.log(`[Migration] localStorage: ${newKey} already exists, skipping`);
+				if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+					console.log(`[Migration] localStorage: ${newKey} already exists, skipping`);
 				continue;
 			}

@@ -127,9 +128,11 @@ const localStorageMigration: Migration = {
 			if (oldValue !== null) {
 				localStorage.setItem(newKey, oldValue);
 				// Keep old key for downgrade compatibility - DO NOT DELETE
-				console.log(
-					`[Migration] localStorage: copied ${deprecatedKey} → ${newKey} (preserved old)`
-				);
+				if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
+					console.log(
+						`[Migration] localStorage: copied ${deprecatedKey} → ${newKey} (preserved old)`
+					);
+				}
 			}
 		}
 	}
@@ -146,7 +149,8 @@ const idxdbMigration: Migration = {
 	async run(): Promise<void> {
 		const oldDbNames = await Dexie.getDatabaseNames();
 		if (!oldDbNames.includes(DB_APP_NAME_DEPRECATED)) {
-			console.log('[Migration] IndexedDB: no old database found, skipping');
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+				console.log('[Migration] IndexedDB: no old database found, skipping');
 			return;
 		}

@@ -155,11 +159,13 @@ const idxdbMigration: Migration = {
 		newDb.version(1).stores(IDXDB_STORES);
 		const existingConvs = await newDb.table(IDXDB_TABLES.conversations).count();
 		if (existingConvs > 0) {
-			console.log('[Migration] IndexedDB: new database already has data, skipping');
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+				console.log('[Migration] IndexedDB: new database already has data, skipping');
 			return;
 		}

-		console.log('[Migration] IndexedDB: copying from', DB_APP_NAME_DEPRECATED);
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log('[Migration] IndexedDB: copying from', DB_APP_NAME_DEPRECATED);

 		const oldDb = new Dexie(DB_APP_NAME_DEPRECATED);
 		oldDb.version(1).stores(IDXDB_STORES);
@@ -169,15 +175,18 @@ const idxdbMigration: Migration = {

 		if (conversations.length > 0) {
 			await newDb.table(IDXDB_TABLES.conversations).bulkAdd(conversations);
-			console.log(`[Migration] IndexedDB: copied ${conversations.length} conversations`);
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+				console.log(`[Migration] IndexedDB: copied ${conversations.length} conversations`);
 		}
 		if (messages.length > 0) {
 			await newDb.table(IDXDB_TABLES.messages).bulkAdd(messages);
-			console.log(`[Migration] IndexedDB: copied ${messages.length} messages`);
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+				console.log(`[Migration] IndexedDB: copied ${messages.length} messages`);
 		}

 		// Non-destructive: DO NOT delete old database - keep for downgrade compatibility
-		console.log('[Migration] IndexedDB: preserved old database for downgrade compatibility');
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log('[Migration] IndexedDB: preserved old database for downgrade compatibility');
 	}
 };

@@ -419,7 +428,8 @@ const legacyMessageMigration: Migration = {
 			}
 		}

-		console.log(`[Migration] Legacy messages: migrated ${migratedCount} messages`);
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log(`[Migration] Legacy messages: migrated ${migratedCount} messages`);
 	}
 };

@@ -434,7 +444,8 @@ const themeMigration: Migration = {
 	async run(): Promise<void> {
 		const legacyTheme = localStorage.getItem('theme');
 		if (legacyTheme === null) {
-			console.log('[Migration] Theme: no legacy theme key found, skipping');
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+				console.log('[Migration] Theme: no legacy theme key found, skipping');
 			return;
 		}

@@ -443,7 +454,8 @@ const themeMigration: Migration = {
 		const config = configRaw ? JSON.parse(configRaw) : {};

 		if (SETTINGS_KEYS.THEME in config) {
-			console.log('[Migration] Theme: config already has theme, skipping');
+			if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+				console.log('[Migration] Theme: config already has theme, skipping');
 			return;
 		}

@@ -451,7 +463,8 @@ const themeMigration: Migration = {
 		localStorage.setItem(CONFIG_LOCALSTORAGE_KEY, JSON.stringify(config));

 		// Non-destructive: DO NOT delete legacy theme key - keep for downgrade compatibility
-		console.log(`[Migration] Theme: copied standalone theme to config (preserved old key)`);
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log(`[Migration] Theme: copied standalone theme to config (preserved old key)`);
 	}
 };

@@ -491,7 +504,8 @@ export const MigrationService = {
 	 */
 	resetState(): void {
 		localStorage.removeItem(MIGRATION_STATE_KEY);
-		console.log('[Migration] State reset - all migrations will run again');
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log('[Migration] State reset - all migrations will run again');
 	},

 	/**
@@ -500,25 +514,30 @@ export const MigrationService = {
 	 */
 	async runAllMigrations(): Promise<void> {
 		const state = getMigrationState();
-		console.log('[Migration] Starting migration run, state:', state);
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log('[Migration] Starting migration run, state:', state);

 		for (const migration of migrations) {
 			if (isMigrationCompleted(migration.id)) {
-				console.log(`[Migration] ${migration.id}: already completed, skipping`);
+				if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+					console.log(`[Migration] ${migration.id}: already completed, skipping`);
 				continue;
 			}

 			try {
-				console.log(`[Migration] ${migration.id}: running...`);
+				if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+					console.log(`[Migration] ${migration.id}: running...`);
 				await migration.run();
 				markMigrationCompleted(migration.id);
-				console.log(`[Migration] ${migration.id}: completed successfully`);
+				if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+					console.log(`[Migration] ${migration.id}: completed successfully`);
 			} catch (error) {
 				console.error(`[Migration] ${migration.id}: failed`, error);
 				markMigrationFailed(migration.id);
 			}
 		}

-		console.log('[Migration] All migrations complete');
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+			console.log('[Migration] All migrations complete');
 	}
 };
@@ -20,11 +20,11 @@
 */

 import { browser } from '$app/environment';
-import { base } from '$app/paths';
 import { SETTINGS_KEYS } from '$lib/constants';
 import { MCPService } from '$lib/services/mcp.service';
 import { config, settingsStore } from '$lib/stores/settings.svelte';
 import { mcpResourceStore } from '$lib/stores/mcp-resources.svelte';
+import { serverStore } from '$lib/stores/server.svelte';
 import { mode } from 'mode-watcher';
 import {
 	parseMcpServerSettings,
@@ -43,7 +43,6 @@ import {
 	ToolCallType
 } from '$lib/enums';
 import {
-	CORS_PROXY_ENDPOINT,
 	DEFAULT_CACHE_TTL_MS,
 	DEFAULT_MCP_CONFIG,
 	EXPECTED_THEMED_ICON_PAIR_COUNT,
@@ -86,7 +85,6 @@ class MCPStore {
 	private _toolCount = $state(0);
 	private _connectedServers = $state<string[]>([]);
 	private _healthChecks = $state<Record<string, HealthCheckState>>({});
-	private _proxyAvailable = $state(false);

 	private connections = new Map<string, MCPConnection>();
 	private toolsIndex = new Map<string, string>();
@@ -96,27 +94,8 @@ class MCPStore {
 	private initPromise: Promise<boolean> | null = null;
 	private activeFlowCount = 0;

-	constructor() {
-		if (browser) {
-			this.probeProxy();
-		}
-	}
-
-	/**
-	 * Probes the CORS proxy endpoint to determine availability.
-	 * The endpoint is only registered when llama-server runs with --ui-mcp-proxy.
-	 */
-	async probeProxy(): Promise<void> {
-		try {
-			const response = await fetch(`${base}${CORS_PROXY_ENDPOINT}`, { method: 'HEAD' });
-			this._proxyAvailable = response.status !== 404;
-		} catch {
-			this._proxyAvailable = false;
-		}
-	}
-
 	get isProxyAvailable(): boolean {
-		return this._proxyAvailable;
+		return serverStore.props?.cors_proxy_enabled ?? false;
 	}

 	/**
@@ -3,7 +3,7 @@ import { toast } from 'svelte-sonner';
 import { ServerModelStatus, ModelModality } from '$lib/enums';
 import { ModelsService } from '$lib/services/models.service';
 import { PropsService } from '$lib/services/props.service';
-import { serverStore } from '$lib/stores/server.svelte';
+import { serverStore, isRouterMode } from '$lib/stores/server.svelte';
 import { TTLCache } from '$lib/utils';
 import {
 	MODEL_PROPS_CACHE_TTL_MS,
@@ -14,14 +14,7 @@ import {
 import { conversationsStore } from '$lib/stores/conversations.svelte';

 /**
- * modelsStore - Reactive store for model management in both MODEL and ROUTER modes
- *
- * This store manages:
- * - Available models list
- * - Selected model for new conversations
- * - Loaded models tracking (ROUTER mode)
- * - Model usage tracking per conversation
- * - Automatic unloading of unused models
+ * modelsStore - Reactive store for model management in both MODEL and ROUTER modes.
 *
 * **Architecture & Relationships:**
 * - **ModelsService**: Stateless service for model API communication
@@ -31,14 +24,8 @@ import { conversationsStore } from '$lib/stores/conversations.svelte';
 *
 * **API Inconsistency Workaround:**
 * In MODEL mode, `/props` returns modalities for the single model.
- * In ROUTER mode, `/props` has no modalities - must use `/props?model=<id>` per model.
+ * In ROUTER mode, `/props` has no modalities — must use `/props?model=<id>` per model.
 * This store normalizes this behavior so consumers don't need to know the server mode.
- *
- * **Key Features:**
- * - **MODEL mode**: Single model, always loaded
- * - **ROUTER mode**: Multi-model with load/unload capability
- * - **Auto-unload**: Automatically unloads models not used by any conversation
- * - **Lazy loading**: ensureModelLoaded() loads models on demand
 */
 class ModelsStore {
 	/**
@@ -57,8 +44,8 @@ class ModelsStore {
 	selectedModelId = $state<string | null>(null);
 	selectedModelName = $state<string | null>(null);

-	// dedup concurrent fetch() callers, all awaiters share the same inflight promise
-	// without this, ?model=<name> URL handler raced an in-progress fetch and saw an empty list
+	// Dedup concurrent fetch() callers — all awaiters share the same inflight promise.
+	// Without this, ?model=<name> URL handler races an in-progress fetch and sees an empty list.
 	private inflightFetch: Promise<void> | null = null;

 	private modelUsage = $state<Map<string, SvelteSet<string>>>(new Map());
@@ -67,9 +54,9 @@ class ModelsStore {
 	favoriteModelIds = $state<Set<string>>(this.loadFavoritesFromStorage());

 	/**
-	 * Model-specific props cache with TTL
-	 * Key: modelId, Value: props data including modalities
-	 * TTL: 10 minutes - props don't change frequently
+	 * Model-specific props cache with TTL.
+	 * Key: modelId, Value: props data including modalities.
+	 * TTL: 10 minutes — props don't change frequently.
 	 */
 	private modelPropsCache = new TTLCache<string, ApiLlamaCppServerProps>({
 		ttlMs: MODEL_PROPS_CACHE_TTL_MS,
@@ -78,7 +65,7 @@ class ModelsStore {
 	private modelPropsFetching = $state<Set<string>>(new Set());

 	/**
-	 * Version counter for props cache - used to trigger reactivity when props are updated
+	 * Version counter for props cache — used to trigger reactivity when props are updated.
 	 */
 	propsCacheVersion = $state(0);

@@ -92,7 +79,7 @@ class ModelsStore {

 	get selectedModel(): ModelOption | null {
 		if (!this.selectedModelId) return null;
-		return this.models.find((model) => model.id === this.selectedModelId) ?? null;
+		return this.models.find((m) => m.id === this.selectedModelId) ?? null;
 	}

 	get loadedModelIds(): string[] {
@@ -117,7 +104,7 @@ class ModelsStore {
 	 * In ROUTER mode, returns null (model is per-conversation).
 	 */
 	get singleModelName(): string | null {
-		if (serverStore.isRouterMode) return null;
+		if (isRouterMode()) return null;

 		const props = serverStore.props;
 		if (props?.model_alias) return props.model_alias;
@@ -126,6 +113,11 @@ class ModelsStore {
 		return props.model_path.split(/(\\|\/)/).pop() || null;
 	}

+	get selectedModelContextSize(): number | null {
+		if (!this.selectedModelName) return null;
+		return this.getModelContextSize(this.selectedModelName);
+	}
+
 	/**
 	 *
 	 *
@@ -134,10 +126,6 @@ class ModelsStore {
 	 *
 	 */

-	/**
-	 * Get modalities for a specific model
-	 * Returns cached modalities from model props
-	 */
 	getModelModalities(modelId: string): ModelModalities | null {
 		const model = this.models.find((m) => m.model === modelId || m.id === modelId);
 		if (model?.modalities) {
@@ -146,46 +134,29 @@ class ModelsStore {

 		const props = this.modelPropsCache.get(modelId);
 		if (props?.modalities) {
-			return {
-				vision: props.modalities.vision ?? false,
-				audio: props.modalities.audio ?? false,
-				video: props.modalities.video ?? false
-			};
+			return this.buildModalities(props.modalities);
 		}

 		return null;
 	}

-	/**
-	 * Check if a model supports vision modality
-	 */
 	modelSupportsVision(modelId: string): boolean {
 		return this.getModelModalities(modelId)?.vision ?? false;
 	}

-	/**
-	 * Check if a model supports audio modality
-	 */
 	modelSupportsAudio(modelId: string): boolean {
 		return this.getModelModalities(modelId)?.audio ?? false;
 	}

-	/**
-	 * Check if a model supports video modality
-	 */
 	modelSupportsVideo(modelId: string): boolean {
 		return this.getModelModalities(modelId)?.video ?? false;
 	}

-	/**
-	 * Get model modalities as an array of ModelModality enum values
-	 */
 	getModelModalitiesArray(modelId: string): ModelModality[] {
 		const modalities = this.getModelModalities(modelId);
 		if (!modalities) return [];

 		const result: ModelModality[] = [];
-
 		if (modalities.vision) result.push(ModelModality.VISION);
 		if (modalities.audio) result.push(ModelModality.AUDIO);
 		if (modalities.video) result.push(ModelModality.VIDEO);
@@ -193,16 +164,10 @@ class ModelsStore {
 		return result;
 	}

-	/**
-	 * Get props for a specific model (from cache)
-	 */
 	getModelProps(modelId: string): ApiLlamaCppServerProps | null {
 		return this.modelPropsCache.get(modelId);
 	}

-	/**
-	 * Get context size (n_ctx) for a specific model from cached props
-	 */
 	getModelContextSize(modelId: string): number | null {
 		const props = this.getModelProps(modelId);
 		const nCtx = props?.default_generation_settings?.n_ctx;
@@ -210,17 +175,6 @@ class ModelsStore {
 		return typeof nCtx === 'number' ? nCtx : null;
 	}

-	/**
-	 * Get context size for the currently selected model or null if no model is selected
-	 */
-	get selectedModelContextSize(): number | null {
-		if (!this.selectedModelName) return null;
-		return this.getModelContextSize(this.selectedModelName);
-	}
-
-	/**
-	 * Check if props are being fetched for a model
-	 */
 	isModelPropsFetching(modelId: string): boolean {
 		return this.modelPropsFetching.has(modelId);
 	}
@@ -235,10 +189,10 @@ class ModelsStore {

 	isModelLoaded(modelId: string): boolean {
 		const model = this.routerModels.find((m) => m.id === modelId);
+
 		return (
 			model?.status.value === ServerModelStatus.LOADED ||
-			model?.status.value === ServerModelStatus.SLEEPING ||
-			false
+			model?.status.value === ServerModelStatus.SLEEPING
 		);
 	}

@@ -248,6 +202,7 @@ class ModelsStore {

 	getModelStatus(modelId: string): ServerModelStatus | null {
 		const model = this.routerModels.find((m) => m.id === modelId);
+
 		return model?.status.value ?? null;
 	}

@@ -257,6 +212,7 @@ class ModelsStore {

 	isModelInUse(modelId: string): boolean {
 		const usage = this.modelUsage.get(modelId);
+
 		return usage !== undefined && usage.size > 0;
 	}

@@ -269,8 +225,8 @@ class ModelsStore {
 	 */

 	/**
-	 * Fetch list of models from server and detect server role
-	 * Also fetches modalities for MODEL mode (single model)
+	 * Fetch list of models from server and detect server role.
+	 * Also fetches modalities for MODEL mode (single model).
 	 */
 	async fetch(force = false): Promise<void> {
 		if (this.inflightFetch) return this.inflightFetch;
@@ -293,69 +249,87 @@ class ModelsStore {
 				await serverStore.fetch();
 			}

-			const response = await ModelsService.list();
+			const router = isRouterMode();

-			const models: ModelOption[] = response.data.map((item: ApiModelDataEntry, index: number) => {
-				const details = response.models?.[index];
-				const rawCapabilities = Array.isArray(details?.capabilities) ? details?.capabilities : [];
-				const displayNameSource =
-					details?.name && details.name.trim().length > 0 ? details.name : item.id;
-				const displayName = this.toDisplayName(displayNameSource);
-				const modelId = details?.model || item.id;
+			if (router) {
+				const response = await ModelsService.listRouter();

-				return {
-					id: item.id,
-					name: displayName,
-					model: modelId,
-					description: details?.description,
-					capabilities: rawCapabilities.filter((value: unknown): value is string => Boolean(value)),
-					details: details?.details,
-					meta: item.meta ?? null,
-					parsedId: ModelsService.parseModelId(modelId),
-					aliases: item.aliases ?? [],
-					tags: item.tags ?? []
-				} satisfies ModelOption;
-			});
+				this.routerModels = response.data;
+				this.models = this.buildModelOptions(response);

-			this.models = models;
+				await this.fetchModalitiesForLoadedModels();

-			// WORKAROUND: In MODEL mode, /props returns modalities for the single model,
-			// but /v1/models doesn't include modalities. We bridge this gap here.
-			const serverProps = serverStore.props;
-			if (serverStore.isModelMode && this.models.length > 0 && serverProps?.modalities) {
-				const modalities: ModelModalities = {
-					vision: serverProps.modalities.vision ?? false,
-					audio: serverProps.modalities.audio ?? false,
-					video: serverProps.modalities.video ?? false
-				};
-				this.modelPropsCache.set(this.models[0].model, serverProps);
-				this.models = this.models.map((model, index) =>
-					index === 0 ? { ...model, modalities } : model
-				);
+				const visible = this.getVisibleModels();
+
+				if (visible.length === 1 && this.isModelLoaded(visible[0].model)) {
+					this.selectModelById(visible[0].id);
+				}
+			} else {
+				this.models = await this.fetchModelModeInternal();
 			}
 		} catch (error) {
 			this.models = [];
 			this.error = error instanceof Error ? error.message : 'Failed to load models';
+
 			throw error;
 		} finally {
 			this.loading = false;
 		}
 	}

+	/** Fetch models in MODEL mode (single model, standard OpenAI-compatible). */
+	private async fetchModelModeInternal(): Promise<ModelOption[]> {
+		const response = await ModelsService.list();
+
+		return this.buildModelOptions(response);
+	}
+
 	/**
-	 * Fetch router models with full metadata (ROUTER mode only)
-	 * This fetches the /models endpoint which returns status info for each model
+	 * Build ModelOption[] from an API response.
+	 * Both MODEL and ROUTER modes share the same mapping logic;
+	 * they differ only in which endpoint is called.
+	 */
+	private buildModelOptions(
+		response: ApiModelListResponse | ApiRouterModelsListResponse
+	): ModelOption[] {
+		return response.data.map((item: ApiModelDataEntry, index: number) => {
+			const details = response.models?.[index];
+			const rawCapabilities = Array.isArray(details?.capabilities) ? details?.capabilities : [];
+			const displayNameSource =
+				details?.name && details.name.trim().length > 0 ? details.name : item.id;
+			const modelId = details?.model || item.id;
+
+			return {
+				id: item.id,
+				name: this.toDisplayName(displayNameSource),
+				model: modelId,
+				description: details?.description,
+				capabilities: rawCapabilities.filter((value: unknown): value is string => Boolean(value)),
+				details: details?.details,
+				meta: item.meta ?? null,
+				parsedId: ModelsService.parseModelId(modelId),
+				aliases: item.aliases ?? [],
+				tags: item.tags ?? []
+			};
+		});
+	}
+
+	/**
+	 * Fetch router models with full metadata (ROUTER mode only).
+	 * No-op in router mode — fetch() already calls listRouter() internally.
+	 * Kept for API compatibility (e.g. handleOpenChange dropdown open handler).
 	 */
 	async fetchRouterModels(): Promise<void> {
+		if (!isRouterMode()) return;
+
 		try {
 			const response = await ModelsService.listRouter();
 			this.routerModels = response.data;
 			await this.fetchModalitiesForLoadedModels();

-			const o = this.models.filter((option) => this.getModelProps(option.model)?.ui !== false);
-
-			if (o.length === 1 && this.isModelLoaded(o[0].model)) {
-				this.selectModelById(o[0].id);
+			const visible = this.getVisibleModels();
+			if (visible.length === 1 && this.isModelLoaded(visible[0].model)) {
+				this.selectModelById(visible[0].id);
 			}
 		} catch (error) {
 			console.warn('Failed to fetch router models:', error);
@@ -364,10 +338,10 @@ class ModelsStore {
 	}

 	/**
-	 * Fetch props for a specific model from /props endpoint
-	 * Uses caching to avoid redundant requests
+	 * Fetch props for a specific model from /props endpoint.
+	 * Uses caching to avoid redundant requests.
 	 *
-	 * In ROUTER mode, this will only fetch props if the model is loaded,
+	 * In ROUTER mode, this only fetches props if the model is loaded,
 	 * since unloaded models return 400 from /props endpoint.
 	 *
 	 * @param modelId - Model identifier to fetch props for
@@ -397,10 +371,7 @@ class ModelsStore {
 		}
 	}

-	/**
-	 * Fetch modalities for all loaded models from /props endpoint
-	 * This updates the modalities field in models array
-	 */
+	/** Fetch modalities for all loaded models from /props endpoint. */
 	async fetchModalitiesForLoadedModels(): Promise<void> {
 		const loadedModelIds = this.loadedModelIds;
 		if (loadedModelIds.length === 0) return;
@@ -410,7 +381,6 @@ class ModelsStore {
 		try {
 			const results = await Promise.all(propsPromises);

-			// Update models with modalities
 			this.models = this.models.map((model) => {
 				const modelIndex = loadedModelIds.indexOf(model.model);
 				if (modelIndex === -1) return model;
@@ -418,13 +388,7 @@ class ModelsStore {
 				const props = results[modelIndex];
 				if (!props?.modalities) return model;

-				const modalities: ModelModalities = {
-					vision: props.modalities.vision ?? false,
-					audio: props.modalities.audio ?? false,
-					video: props.modalities.video ?? false
-				};
-
-				return { ...model, modalities };
+				return { ...model, modalities: this.buildModalities(props.modalities) };
 			});

 			this.propsCacheVersion++;
@@ -433,17 +397,38 @@ class ModelsStore {
 		}
 	}

+	/**
+	 * Update modalities for a specific model.
+	 * Called when a model is loaded or when we need fresh modality data.
+	 */
+	async updateModelModalities(modelId: string): Promise<void> {
+		const props = await this.fetchModelProps(modelId);
+		if (!props?.modalities) return;
+
+		this.models = this.models.map((model) =>
+			model.model === modelId
+				? { ...model, modalities: this.buildModalities(props.modalities!) }
+				: model
+		);
+
+		this.propsCacheVersion++;
+	}
+
+	/**
+	 * Filter to models visible in the UI (ui !== false).
+	 */
+	private getVisibleModels(): ModelOption[] {
+		return this.models.filter((option) => this.getModelProps(option.model)?.ui !== false);
+	}
+
 	/**
 	 * Gets the model name from the last assistant message in the active conversation.
-	 * Iterates backward through messages to find the most recent message with a model.
 	 * Used by both the chat page and settings page to maintain model consistency.
-	 * @returns The model name or null if not found
 	 */
 	getModelFromLastAssistantResponse(): string | null {
 		const messages = conversationsStore.activeMessages;
 		if (!messages || messages.length === 0) return null;

-		// Iterate backward to find the last message with a model
 		for (let i = messages.length - 1; i >= 0; i--) {
 			if (messages[i].model) {
 				return messages[i].model;
@@ -456,22 +441,13 @@ class ModelsStore {
 	/**
 	 * Auto-selects the model from the last assistant response if available and loaded.
 	 * Returns true if a model was selected, false otherwise.
-	 * This is used by the chat page to maintain model consistency across page navigation.
 	 */
 	async selectModelFromLastAssistantResponse(): Promise<boolean> {
 		const lastModel = this.getModelFromLastAssistantResponse();
-		if (!lastModel) return false;
-
-		// Skip if already selected
-		if (this.selectedModelName === lastModel) return false;
+		if (!lastModel || this.selectedModelName === lastModel) return false;

 		const matchingModel = this.models.find((option) => option.model === lastModel);
-		if (!matchingModel) return false;
-
-		if (!this.isModelLoaded(lastModel)) {
-			console.log('[modelsStore] last assistant model not loaded:', lastModel);
-			return false;
-		}
+		if (!matchingModel || !this.isModelLoaded(lastModel)) return false;

 		try {
 			await this.selectModelById(matchingModel.id);
@@ -484,22 +460,17 @@ class ModelsStore {
 	}

 	/**
-	 * Auto-selects the first available model if none is selected, and fetches its props.
+	 * Auto-selects the first available model if none is selected.
 	 * Prioritizes:
 	 * 1. Model from active conversation's last assistant response (if loaded)
 	 * 2. Model from active conversation's last assistant response (if not loaded)
 	 * 3. First loaded model (not from active conversation)
 	 * 4. First available model
-	 * This is used to ensure default values are populated in settings pages.
 	 */
 	async ensureFirstModelSelected(): Promise<void> {
 		if (this.selectedModelName) return;

-		// Filter models that are visible in the UI
-		const availableModels = this.models.filter(
-			(option) => this.getModelProps(option.model)?.ui !== false
-		);
-
+		const availableModels = this.getVisibleModels();
 		if (availableModels.length === 0) return;

 		// Try to select model from last assistant response first
@@ -515,7 +486,7 @@ class ModelsStore {
 			}
 		}

-		// Try to find a loaded model first
+		// Try a loaded model first
 		const loadedModel = availableModels.find((m) => this.isModelLoaded(m.model));
 		if (loadedModel) {
 			await this.selectModelById(loadedModel.id);
@@ -524,34 +495,7 @@ class ModelsStore {
 		}

 		// Fall back to the first available model
-		const firstModel = availableModels[0];
-		await this.selectModelById(firstModel.id);
-		// Don't fetch props for unloaded models (will fail in ROUTER mode)
-	}
-
-	/**
-	 * Update modalities for a specific model
-	 * Called when a model is loaded or when we need fresh modality data
-	 */
-	async updateModelModalities(modelId: string): Promise<void> {
-		try {
-			const props = await this.fetchModelProps(modelId);
-			if (!props?.modalities) return;
-
-			const modalities: ModelModalities = {
-				vision: props.modalities.vision ?? false,
-				audio: props.modalities.audio ?? false,
-				video: props.modalities.video ?? false
-			};
-
-			this.models = this.models.map((model) =>
-				model.model === modelId ? { ...model, modalities } : model
-			);
-
-			this.propsCacheVersion++;
-		} catch (error) {
-			console.warn(`Failed to update modalities for model ${modelId}:`, error);
-		}
+		await this.selectModelById(availableModels[0].id);
 	}

 	/**
@@ -562,9 +506,6 @@ class ModelsStore {
 	 *
 	 */

-	/**
-	 * Select a model for new conversations
-	 */
 	async selectModelById(modelId: string): Promise<void> {
 		if (!modelId || this.updating) return;
 		if (this.selectedModelId === modelId) return;
@@ -584,8 +525,7 @@ class ModelsStore {
 	}

 	/**
-	 * Select a model by its model name (used for syncing with conversation model)
-	 * @param modelName - Model name to select (e.g., "ggml-org/GLM-4.7-Flash-GGUF")
+	 * Select a model by its model name (used for syncing with conversation model).
 	 */
 	selectModelByName(modelName: string): void {
 		const option = this.models.find((model) => model.model === modelName);
@@ -615,7 +555,7 @@ class ModelsStore {
 	/**
 	 *
 	 *
-	 * Loading/Unloading Models
+	 * Loading / Unloading Models
 	 *
 	 *
 	 */
@@ -623,27 +563,18 @@ class ModelsStore {
 	/**
 	 * WORKAROUND: Polling for model status after load/unload operations.
 	 *
-	 * Currently, the `/models/load` and `/models/unload` endpoints return success
-	 * before the operation actually completes on the server. This means an immediate
-	 * request to `/models` returns stale status (e.g., "loading" after load request,
-	 * "loaded" after unload request).
+	 * Currently, `/models/load` and `/models/unload` return success before
+	 * the operation actually completes on the server.
 	 *
-	 * TODO: Remove this polling once llama-server properly waits for the operation
-	 * to complete before returning success from `/load` and `/unload` endpoints.
-	 * At that point, a single `fetchRouterModels()` call after the operation will
-	 * be sufficient to get the correct status.
+	 * TODO: Remove polling once llama-server properly waits for the operation
+	 * to complete before returning success.
 	 */

-	/** Polling interval in ms for checking model status */
 	private static readonly STATUS_POLL_INTERVAL = 500;

 	/**
 	 * Poll for expected model status after load/unload operation.
-	 * Keeps polling indefinitely until the model reaches the expected status or fails.
-	 *
-	 * @param modelId - Model identifier to check
-	 * @param expectedStatus - Expected status to wait for
-	 * @throws Error if model reaches FAILED status
+	 * Keeps polling until the model reaches the expected status or fails.
 	 */
 	private async pollForModelStatus(
 		modelId: string,
@@ -654,9 +585,7 @@ class ModelsStore {
 			await this.fetchRouterModels();

 			const currentStatus = this.getModelStatus(modelId);
-			if (currentStatus === expectedStatus) {
-				return;
-			}
+			if (currentStatus === expectedStatus) return;

 			if (currentStatus === ServerModelStatus.FAILED) {
 				throw new Error(
@@ -677,15 +606,8 @@ class ModelsStore {
 		}
 	}

-	/**
-	 * Load a model (ROUTER mode)
-	 * @param modelId - Model identifier to load
-	 */
 	async loadModel(modelId: string): Promise<void> {
-		if (this.isModelLoaded(modelId)) {
-			return;
-		}
-
+		if (this.isModelLoaded(modelId)) return;
 		if (this.modelLoadingStates.get(modelId)) return;

 		this.modelLoadingStates.set(modelId, true);
@@ -694,7 +616,6 @@ class ModelsStore {
 		try {
 			await ModelsService.load(modelId);
 			await this.pollForModelStatus(modelId, ServerModelStatus.LOADED);
-
 			await this.updateModelModalities(modelId);
 			toast.success(`Model loaded: ${this.toDisplayName(modelId)}`);
 		} catch (error) {
@@ -706,15 +627,8 @@ class ModelsStore {
 		}
 	}

-	/**
-	 * Unload a model (ROUTER mode)
-	 * @param modelId - Model identifier to unload
-	 */
 	async unloadModel(modelId: string): Promise<void> {
-		if (!this.isModelLoaded(modelId)) {
-			return;
-		}
-
+		if (!this.isModelLoaded(modelId)) return;
 		if (this.modelLoadingStates.get(modelId)) return;

 		this.modelLoadingStates.set(modelId, true);
@@ -722,7 +636,6 @@ class ModelsStore {

 		try {
 			await ModelsService.unload(modelId);
-
 			await this.pollForModelStatus(modelId, ServerModelStatus.UNLOADED);
 			toast.info(`Model unloaded: ${this.toDisplayName(modelId)}`);
 		} catch (error) {
@@ -734,15 +647,8 @@ class ModelsStore {
 		}
 	}

-	/**
-	 * Ensure a model is loaded before use
-	 * @param modelId - Model identifier to ensure is loaded
-	 */
 	async ensureModelLoaded(modelId: string): Promise<void> {
-		if (this.isModelLoaded(modelId)) {
-			return;
-		}
-
+		if (this.isModelLoaded(modelId)) return;
 		await this.loadModel(modelId);
 	}

@@ -779,11 +685,9 @@ class ModelsStore {
 	private loadFavoritesFromStorage(): Set<string> {
 		try {
 			const raw = localStorage.getItem(FAVORITE_MODELS_LOCALSTORAGE_KEY);
-
 			return raw ? new Set(JSON.parse(raw) as string[]) : new Set();
 		} catch {
 			toast.error('Failed to load favorite models from local storage');
-
 			return new Set();
 		}
 	}
@@ -799,10 +703,19 @@ class ModelsStore {
 	private toDisplayName(id: string): string {
 		const segments = id.split(/\\|\//);
 		const candidate = segments.pop();
-
 		return candidate && candidate.trim().length > 0 ? candidate : id;
 	}

+	private buildModalities(
+		modalities: NonNullable<ApiLlamaCppServerProps['modalities']>
+	): ModelModalities {
+		return {
+			vision: modalities.vision ?? false,
+			audio: modalities.audio ?? false,
+			video: modalities.video ?? false
+		};
+	}
+
 	clear(): void {
 		this.models = [];
 		this.routerModels = [];
@@ -203,6 +203,7 @@ export interface ApiLlamaCppServerProps {
 	/** @deprecated Use {@link ui_settings} instead */
 	webui_settings?: Record<string, string | number | boolean>;
 	ui_settings?: Record<string, string | number | boolean>;
+	cors_proxy_enabled?: boolean;
 }

 export interface ApiChatCompletionRequest {
@@ -1,5 +1,5 @@
-import type { MCPConnectionPhase, MCPLogLevel, HealthCheckStatus } from '$lib/enums/mcp';
-import type { ToolSource } from '$lib/enums/tools';
+import type { MCPConnectionPhase, MCPLogLevel, HealthCheckStatus } from '$lib/enums/mcp.enums';
+import type { ToolSource } from '$lib/enums/tools.enums';
 import type {
 	Client,
 	ClientCapabilities as SDKClientCapabilities,
@@ -12,17 +12,21 @@ export async function validateApiKey(fetch: typeof globalThis.fetch): Promise<vo
 		return;
 	}

+	const apiKey = config().apiKey;
+
+	// No API key configured — server doesn't require auth, skip the request entirely.
+	// The /props endpoint is only protected when the server has API keys configured,
+	// and in that case the client always has one set (from settings).
+	if (!apiKey) {
+		return;
+	}
+
 	try {
-		const apiKey = config().apiKey;
-
 		const headers: Record<string, string> = {
-			'Content-Type': 'application/json'
+			'Content-Type': 'application/json',
+			Authorization: `Bearer ${apiKey}`
 		};

-		if (apiKey) {
-			headers.Authorization = `Bearer ${apiKey}`;
-		}
-
 		const response = await fetch(`${base}/props`, { headers });

 		if (!response.ok) {
@@ -333,7 +333,8 @@ async function migrateConversation(convId: string): Promise<number> {
 export async function runLegacyMigration(): Promise<void> {
 	if (!isMigrationNeeded()) return;

-	console.log('[Migration] Starting legacy message format migration...');
+	if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
+		console.log('[Migration] Starting legacy message format migration...');

 	try {
 		const conversations = await DatabaseService.getAllConversations();
@@ -344,12 +345,14 @@ export async function runLegacyMigration(): Promise<void> {
 			totalMigrated += count;
 		}

-		if (totalMigrated > 0) {
-			console.log(
-				`[Migration] Migrated ${totalMigrated} messages across ${conversations.length} conversations`
-			);
-		} else {
-			console.log('[Migration] No legacy messages found, marking as done');
+		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
+			if (totalMigrated > 0) {
+				console.log(
+					`[Migration] Migrated ${totalMigrated} messages across ${conversations.length} conversations`
+				);
+			} else {
+				console.log('[Migration] No legacy messages found, marking as done');
+			}
 		}

 		markMigrationDone();
@@ -3,7 +3,6 @@
 	import { chatStore } from '$lib/stores/chat.svelte';
 	import { conversationsStore, isConversationsInitialized } from '$lib/stores/conversations.svelte';
 	import { modelsStore, modelOptions } from '$lib/stores/models.svelte';
-	import { isRouterMode } from '$lib/stores/server.svelte';
 	import { onMount } from 'svelte';
 	import { page } from '$app/state';
 	import { replaceState } from '$app/navigation';
@@ -72,23 +71,13 @@
 		conversationsStore.clearActiveConversation();
 		chatStore.clearUIState();

-		if (
-			isRouterMode() &&
-			modelsStore.selectedModelName &&
-			!modelsStore.isModelLoaded(modelsStore.selectedModelName)
-		) {
-			modelsStore.clearSelection();
+		await modelsStore.fetch();

-			const first = modelOptions().find((m) => modelsStore.loadedModelIds.includes(m.model));
-			if (first) {
-				await modelsStore.selectModelById(first.id);
-			}
-		}
-
-		// Handle URL params only if we have ?q= or ?model= or ?new_chat=true
 		if (qParam !== null || modelParam !== null || newChatParam === 'true') {
 			await handleUrlParams();
 		}
+
+		await modelsStore.ensureFirstModelSelected();
 	});
 </script>

@@ -7,11 +7,13 @@
 	import { untrack } from 'svelte';
 	import { onMount } from 'svelte';
 	import { fade } from 'svelte/transition';
+
 	import {
 		DesktopIconStrip,
 		DialogConversationTitleUpdate,
 		SidebarNavigation
 	} from '$lib/components/app';
+
 	import { conversationsStore } from '$lib/stores/conversations.svelte';
 	import * as Sidebar from '$lib/components/ui/sidebar/index.js';
 	import * as Tooltip from '$lib/components/ui/tooltip';
@@ -30,26 +32,29 @@
 	import { conversations } from '$lib/stores/conversations.svelte';

 	let { children } = $props();
-
 	let alwaysShowSidebarOnDesktop = $derived(config().alwaysShowSidebarOnDesktop);
 	let isMobile = new IsMobile();
 	let isDesktop = $derived(!isMobile.current);
 	let sidebarOpen = $state(false);
 	let mounted = $state(false);
 	let innerHeight = $state<number | undefined>();
+
 	let chatSidebar:
-		| { activateSearchMode?: () => void; editActiveConversation?: () => void }
+		| {
+				activateSearchMode?: () => void;
+				editActiveConversation?: () => void;
+		  }
 		| undefined = $state();

 	let titleUpdateDialogOpen = $state(false);
 	let titleUpdateCurrentTitle = $state('');
 	let titleUpdateNewTitle = $state('');
 	let titleUpdateResolve: ((value: boolean) => void) | null = null;
-
 	const panelNav = useSettingsNavigation();

 	function navigateToConversation(direction: -1 | 1) {
 		const allConvs = conversations();
+
 		if (allConvs.length === 0) return;

 		const currentId = page.params.id;
@@ -61,6 +66,7 @@
 		}

 		const idx = allConvs.findIndex((c) => c.id === currentId);
+
 		if (idx === -1) return;

 		const targetIdx = idx + direction;
@@ -75,38 +81,41 @@
 	// Global keyboard shortcuts
 	const { handleKeydown } = useKeyboardShortcuts({
 		editActiveConversation: () => chatSidebar?.editActiveConversation?.(),
-
 		navigateToPrevConversation: () => navigateToConversation(-1),
-
 		navigateToNextConversation: () => navigateToConversation(1)
 	});

 	function checkApiKey() {
 		const apiKey = config().apiKey;

-		if (
-			(page.route.id === '/(chat)' || page.route.id === '/(chat)/chat/[id]') &&
-			page.status !== 401 &&
-			page.status !== 403
-		) {
-			const headers: Record<string, string> = {
-				'Content-Type': 'application/json'
-			};
-
-			if (apiKey && apiKey.trim() !== '') {
-				headers.Authorization = `Bearer ${apiKey.trim()}`;
-			}
-
-			fetch(`${base}/props`, { headers })
-				.then((response) => {
-					if (response.status === 401 || response.status === 403) {
-						window.location.reload();
-					}
-				})
-				.catch((e) => {
-					console.error('Error checking API key:', e);
-				});
+		// No API key configured — server doesn't require auth, no need to validate.
+		// This mirrors the early return in validateApiKey() to avoid redundant /props requests.
+		if (!apiKey || apiKey.trim() === '') {
+			return;
 		}
+
+		untrack(() => {
+			if (
+				(page.route.id === '/(chat)' || page.route.id === '/(chat)/chat/[id]') &&
+				page.status !== 401 &&
+				page.status !== 403
+			) {
+				const headers: Record<string, string> = {
+					'Content-Type': 'application/json',
+					Authorization: `Bearer ${apiKey.trim()}`
+				};
+
+				fetch(`${base}/props`, { headers })
+					.then((response) => {
+						if (response.status === 401 || response.status === 403) {
+							window.location.reload();
+						}
+					})
+					.catch((e) => {
+						console.error('Error checking API key:', e);
+					});
+			}
+		});
 	}

 	function handleTitleUpdateCancel() {
@@ -134,6 +143,7 @@
 	$effect(() => {
 		if (alwaysShowSidebarOnDesktop && isDesktop) {
 			sidebarOpen = true;
+
 			return;
 		}
 	});
@@ -170,6 +180,7 @@
 		// Only fetch router models once when we have models loaded and in router mode
 		if (isRouter && modelsCount > 0 && !routerModelsFetched) {
 			routerModelsFetched = true;
+
 			untrack(() => {
 				modelsStore.fetchRouterModels();
 			});
@@ -218,7 +229,6 @@

 <Tooltip.Provider delayDuration={TOOLTIP_DELAY_DURATION}>
 	<ModeWatcher />
-
 	<Toaster richColors />

 	<DialogConversationTitleUpdate
@@ -230,10 +240,10 @@
 	/>

 	<Sidebar.Provider bind:open={sidebarOpen}>
-		<div class="flex h-screen w-full" style:height="{innerHeight}px">
-			<Sidebar.Root variant="floating" class="h-full">
-				<SidebarNavigation bind:this={chatSidebar} />
-			</Sidebar.Root>
+		<div class="flex h-screen w-full">
+			<Sidebar.Root variant="floating" class="h-full"
+				><SidebarNavigation bind:this={chatSidebar} /></Sidebar.Root
+			>

 			{#if !(alwaysShowSidebarOnDesktop && isDesktop) && !(panelNav.isSettingsRoute && !isDesktop)}
 				{#if mounted}
@@ -261,9 +271,9 @@
 				/>
 			{/if}

-			<Sidebar.Inset class="flex flex-1 flex-col overflow-hidden">
-				{@render children?.()}
-			</Sidebar.Inset>
+			<Sidebar.Inset class="flex flex-1 flex-col overflow-hidden"
+				>{@render children?.()}</Sidebar.Inset
+			>
 		</div>
 	</Sidebar.Provider>
 </Tooltip.Provider>
@@ -8,6 +8,9 @@ $use-ttf: false;
 $font-folder: 'katex-fonts';

 // Import KaTeX SCSS with overridden variables
-// Note: @import is deprecated but required because KaTeX uses @import internally
-// The deprecation warnings are from KaTeX's code and cannot be avoided
-@import 'katex/src/styles/katex.scss';
+@use 'katex/src/styles/katex.scss' with (
+	$use-woff2: true,
+	$use-woff: false,
+	$use-ttf: false,
+	$font-folder: 'katex-fonts'
+);
@@ -4,8 +4,9 @@ import TestWrapper from './components/TestWrapper.svelte';

 describe('/+page.svelte', () => {
 	it('should render page without throwing', async () => {
-		// Basic smoke test - page should render without throwing errors
-		// API calls will fail in test environment but component should still mount
-		expect(() => render(TestWrapper)).not.toThrow();
+		// Basic smoke test - page should render without throwing errors.
+		// API calls are mocked in vitest-setup-client.ts.
+		await render(TestWrapper);
+		expect(true).toBe(true);
 	});
 });
@@ -23,18 +23,6 @@ export default defineConfig({
 		minify: true
 	},

-	css: {
-		preprocessorOptions: {
-			scss: {
-				additionalData: `
-					$use-woff2: true;
-					$use-woff: false;
-					$use-ttf: false;
-				`
-			}
-		}
-	},
-
 	plugins: [tailwindcss(), sveltekit(), devtoolsJson(), llamaCppBuildPlugin()],

 	test: {
@@ -1,2 +1,80 @@
 /// <reference types="@vitest/browser/matchers" />
 /// <reference types="@vitest/browser/providers/playwright" />
+
+import { beforeEach, vi } from 'vitest';
+
+// Mock fetch for API calls during client tests.
+// In test environment there is no backend server, so we intercept
+// the specific endpoints the app uses and return valid mock data.
+beforeEach(() => {
+	const originalFetch = globalThis.fetch;
+
+	vi.spyOn(globalThis, 'fetch').mockImplementation(
+		async (input: RequestInfo | URL, init?: RequestInit) => {
+			const url = typeof input === 'string' ? input : input instanceof URL ? input.href : input.url;
+
+			// Mock server props endpoint
+			if (url.includes('/server')) {
+				return new Response(
+					JSON.stringify({
+						mode: 'router',
+						version: 'test',
+						git_commit: 'test',
+						git_branch: 'test'
+					}),
+					{ status: 200, headers: { 'Content-Type': 'application/json' } }
+				);
+			}
+
+			// Mock models list endpoint
+			if (/\/v1\/models|\/models\b/.test(url)) {
+				return new Response(
+					JSON.stringify({
+						object: 'list',
+						data: [
+							{
+								id: 'test-model.gguf',
+								object: 'model',
+								owned_by: 'llamacpp',
+								created: 0,
+								in_cache: false,
+								path: 'models/test-model.gguf',
+								status: { value: 'unloaded' },
+								meta: {}
+							}
+						],
+						models: [
+							{
+								model: 'test-model.gguf',
+								name: 'Test Model',
+								details: {}
+							}
+						]
+					}),
+					{ status: 200, headers: { 'Content-Type': 'application/json' } }
+				);
+			}
+
+			// Mock /props endpoint (used for modalities)
+			if (url.includes('/props')) {
+				return new Response(
+					JSON.stringify({
+						default_generation_settings: { n_ctx: 2048 }
+					}),
+					{ status: 200, headers: { 'Content-Type': 'application/json' } }
+				);
+			}
+
+			// Mock /tools endpoint (used for built-in tools list)
+			if (url.includes('/tools')) {
+				return new Response(JSON.stringify([]), {
+					status: 200,
+					headers: { 'Content-Type': 'application/json' }
+				});
+			}
+
+			// Default: use real fetch
+			return originalFetch(input, init);
+		}
+	);
+});
@@ -0,0 +1 @@
+/// <reference types="@vitest/browser-playwright" />
				`@@ -0,0 +1 @@`
				`/// <reference types="@vitest/browser-playwright" />`