diff --git a/internal/config/model_config.go b/internal/config/model_config.go
index 7e354779..b85d7ce1 100644
--- a/internal/config/model_config.go
+++ b/internal/config/model_config.go
@@ -14,6 +14,7 @@ var validModalities = map[string]struct{}{
 	"text":  {},
 	"audio": {},
 	"image": {},
+	"video": {},
 }
 
 // ModelCapConfig defines what modalities and features a model supports.
@@ -37,12 +38,12 @@ func (c ModelCapConfig) Empty() bool {
 func (c ModelCapConfig) Validate() error {
 	for _, m := range c.In {
 		if _, ok := validModalities[m]; !ok {
-			return fmt.Errorf("capabilities.in: invalid modality %q, must be one of: text, audio, image", m)
+			return fmt.Errorf("capabilities.in: invalid modality %q, must be one of: text, audio, image, video", m)
 		}
 	}
 	for _, m := range c.Out {
 		if _, ok := validModalities[m]; !ok {
-			return fmt.Errorf("capabilities.out: invalid modality %q, must be one of: text, audio, image", m)
+			return fmt.Errorf("capabilities.out: invalid modality %q, must be one of: text, audio, image, video", m)
 		}
 	}
 	if c.Context < 0 {
diff --git a/internal/config/model_config_test.go b/internal/config/model_config_test.go
index 622dab4e..eb72b4fd 100644
--- a/internal/config/model_config_test.go
+++ b/internal/config/model_config_test.go
@@ -296,20 +296,25 @@ func TestConfig_ModelCapabilities_Validate(t *testing.T) {
 		assert.NoError(t, caps.Validate())
 	})
 
+	t.Run("valid_video_modality", func(t *testing.T) {
+		caps := ModelCapConfig{In: []string{"text", "image"}, Out: []string{"video"}}
+		assert.NoError(t, caps.Validate())
+	})
+
 	t.Run("invalid_in_modality", func(t *testing.T) {
-		caps := ModelCapConfig{In: []string{"video"}}
+		caps := ModelCapConfig{In: []string{"smell"}}
 		err := caps.Validate()
 		assert.Error(t, err)
 		assert.Contains(t, err.Error(), "capabilities.in")
-		assert.Contains(t, err.Error(), "video")
+		assert.Contains(t, err.Error(), "smell")
 	})
 
 	t.Run("invalid_out_modality", func(t *testing.T) {
-		caps := ModelCapConfig{Out: []string{"video"}}
+		caps := ModelCapConfig{Out: []string{"smell"}}
 		err := caps.Validate()
 		assert.Error(t, err)
 		assert.Contains(t, err.Error(), "capabilities.out")
-		assert.Contains(t, err.Error(), "video")
+		assert.Contains(t, err.Error(), "smell")
 	})
 
 	t.Run("negative_context", func(t *testing.T) {
@@ -327,10 +332,10 @@ models:
     capabilities:
       in:
         - text
-        - video
+        - smell
 `
 		_, err := LoadConfigFromReader(strings.NewReader(content))
 		assert.Error(t, err)
-		assert.Contains(t, err.Error(), "video")
+		assert.Contains(t, err.Error(), "smell")
 	})
 }
diff --git a/internal/server/api.go b/internal/server/api.go
index 0da4a781..b1b4f107 100644
--- a/internal/server/api.go
+++ b/internal/server/api.go
@@ -88,6 +88,12 @@ func renderCapabilities(caps config.ModelCapConfig) (arch map[string]any, capsMa
 		if contains(caps.In, "image") && contains(caps.Out, "image") {
 			capsMap["image_to_image"] = true
 		}
+		if contains(caps.In, "text") && contains(caps.Out, "video") {
+			capsMap["video_generation"] = true
+		}
+		if contains(caps.In, "image") && contains(caps.Out, "video") {
+			capsMap["image_to_video"] = true
+		}
 	}
 
 	if caps.Tools {
diff --git a/ui-svelte/src/components/playground/VideoInterface.svelte b/ui-svelte/src/components/playground/VideoInterface.svelte
new file mode 100644
index 00000000..1181bd8a
--- /dev/null
+++ b/ui-svelte/src/components/playground/VideoInterface.svelte
@@ -0,0 +1,359 @@
+<script lang="ts">
+  import { models } from "../../stores/api";
+  import { persistentStore } from "../../stores/persistent";
+  import { generateVideoSync } from "../../lib/videoApi";
+  import { playgroundStores } from "../../stores/playgroundActivity";
+  import ModelSelector from "./ModelSelector.svelte";
+  import ExpandableTextarea from "./ExpandableTextarea.svelte";
+
+  const selectedModelStore = persistentStore<string>("playground-video-model", "");
+  const selectedSizeStore = persistentStore<string>("playground-video-size", "");
+  const negativePromptStore = persistentStore<string>("playground-video-negative-prompt", "");
+  const numFramesStore = persistentStore<number>("playground-video-num-frames", 0);
+  const fpsStore = persistentStore<number>("playground-video-fps", 0);
+  const stepsStore = persistentStore<number>("playground-video-steps", 0);
+  const guidanceStore = persistentStore<number>("playground-video-guidance", 0);
+  const seedStore = persistentStore<number>("playground-video-seed", -1);
+
+  let prompt = $state("");
+  let isGenerating = $state(false);
+  let videoUrl = $state<string | null>(null);
+  let videoMime = $state("video/mp4");
+  let videoBytes = $state(0);
+  let elapsedSeconds = $state(0);
+  let error = $state<string | null>(null);
+  let abortController = $state<AbortController | null>(null);
+  let showSettings = $state(false);
+  let initImage = $state<File | null>(null);
+  let initImagePreview = $state<string | null>(null);
+  let fileInput = $state<HTMLInputElement | null>(null);
+  let timer: ReturnType<typeof setInterval> | null = null;
+
+  let hasModels = $derived($models.some((m) => !m.unlisted));
+
+  $effect(() => {
+    playgroundStores.videoGenerating.set(isGenerating);
+  });
+
+  function onInitImageChange(event: Event) {
+    const input = event.target as HTMLInputElement;
+    const file = input.files?.[0] ?? null;
+    setInitImage(file);
+  }
+
+  function setInitImage(file: File | null) {
+    if (initImagePreview) URL.revokeObjectURL(initImagePreview);
+    initImage = file;
+    initImagePreview = file ? URL.createObjectURL(file) : null;
+  }
+
+  function clearInitImage() {
+    setInitImage(null);
+    if (fileInput) fileInput.value = "";
+  }
+
+  async function generate() {
+    const trimmedPrompt = prompt.trim();
+    if (!trimmedPrompt || !$selectedModelStore || isGenerating) return;
+
+    isGenerating = true;
+    error = null;
+    elapsedSeconds = 0;
+    abortController = new AbortController();
+    timer = setInterval(() => (elapsedSeconds += 1), 1000);
+
+    try {
+      const result = await generateVideoSync(
+        $selectedModelStore,
+        trimmedPrompt,
+        {
+          negativePrompt: $negativePromptStore || undefined,
+          size: $selectedSizeStore || undefined,
+          numFrames: $numFramesStore,
+          fps: $fpsStore,
+          steps: $stepsStore,
+          guidanceScale: $guidanceStore,
+          seed: $seedStore,
+          initImage,
+        },
+        abortController.signal
+      );
+      if (videoUrl) URL.revokeObjectURL(videoUrl);
+      videoUrl = result.url;
+      videoMime = result.mime;
+      videoBytes = result.sizeBytes;
+    } catch (err) {
+      if (err instanceof Error && err.name === "AbortError") {
+        // User cancelled
+      } else {
+        error = err instanceof Error ? err.message : "An error occurred";
+      }
+    } finally {
+      isGenerating = false;
+      abortController = null;
+      if (timer) clearInterval(timer);
+      timer = null;
+    }
+  }
+
+  function cancelGeneration() {
+    abortController?.abort();
+  }
+
+  function clearVideo() {
+    if (videoUrl) URL.revokeObjectURL(videoUrl);
+    videoUrl = null;
+    error = null;
+    prompt = "";
+  }
+
+  function downloadVideo() {
+    if (!videoUrl) return;
+    const ext = videoMime.includes("webm") ? "webm" : "mp4";
+    const link = document.createElement("a");
+    link.href = videoUrl;
+    link.download = `generated-video-${Date.now()}.${ext}`;
+    document.body.appendChild(link);
+    link.click();
+    document.body.removeChild(link);
+  }
+
+  function formatBytes(n: number): string {
+    if (n > 1024 * 1024) return `${(n / 1024 / 1024).toFixed(1)} MB`;
+    if (n > 1024) return `${(n / 1024).toFixed(0)} KB`;
+    return `${n} B`;
+  }
+
+  function formatElapsed(s: number): string {
+    const m = Math.floor(s / 60);
+    return m > 0 ? `${m}m ${s % 60}s` : `${s}s`;
+  }
+
+  function handleKeyDown(event: KeyboardEvent) {
+    if (event.key === "Enter" && !event.shiftKey) {
+      event.preventDefault();
+      generate();
+    }
+  }
+</script>
+
+<div class="flex flex-col h-full">
+  <!-- Model selector and options -->
+  <div class="shrink-0 flex flex-wrap gap-2 mb-4">
+    <ModelSelector
+      bind:value={$selectedModelStore}
+      placeholder="Select a video model..."
+      disabled={isGenerating}
+      capabilities={["video_generation", "image_to_video"]}
+      matchAny={true}
+    />
+
+    <select
+      class="px-3 py-2 rounded border border-gray-200 dark:border-white/10 bg-surface focus:outline-none focus:ring-2 focus:ring-primary"
+      bind:value={$selectedSizeStore}
+      disabled={isGenerating}
+    >
+      <option value="">Model default size</option>
+      <optgroup label="Landscape">
+        <option value="1280x704">1280x704</option>
+        <option value="832x480">832x480</option>
+      </optgroup>
+      <optgroup label="Portrait">
+        <option value="704x1280">704x1280</option>
+        <option value="480x832">480x832</option>
+      </optgroup>
+      <optgroup label="Square">
+        <option value="704x704">704x704</option>
+      </optgroup>
+    </select>
+
+    <button
+      class="px-3 py-2 rounded border border-gray-200 dark:border-white/10 bg-surface hover:bg-secondary-hover transition-colors"
+      onclick={() => (showSettings = !showSettings)}
+    >
+      {showSettings ? "Hide Settings" : "Settings"}
+    </button>
+  </div>
+
+  <!-- Settings panel -->
+  {#if showSettings}
+    <div class="shrink-0 mb-4 p-4 rounded border border-gray-200 dark:border-white/10 bg-surface">
+      <div class="grid grid-cols-2 md:grid-cols-5 gap-3 mb-3">
+        <label class="flex flex-col gap-1">
+          <span class="text-xs text-txtsecondary">Frames (0 = default)</span>
+          <input
+            type="number"
+            class="px-2 py-1 rounded border border-gray-200 dark:border-white/10 bg-surface focus:outline-none focus:ring-2 focus:ring-primary"
+            bind:value={$numFramesStore}
+            min="0"
+            max="1000"
+          />
+        </label>
+        <label class="flex flex-col gap-1">
+          <span class="text-xs text-txtsecondary">FPS (0 = default)</span>
+          <input
+            type="number"
+            class="px-2 py-1 rounded border border-gray-200 dark:border-white/10 bg-surface focus:outline-none focus:ring-2 focus:ring-primary"
+            bind:value={$fpsStore}
+            min="0"
+            max="60"
+          />
+        </label>
+        <label class="flex flex-col gap-1">
+          <span class="text-xs text-txtsecondary">Steps (0 = default)</span>
+          <input
+            type="number"
+            class="px-2 py-1 rounded border border-gray-200 dark:border-white/10 bg-surface focus:outline-none focus:ring-2 focus:ring-primary"
+            bind:value={$stepsStore}
+            min="0"
+            max="150"
+          />
+        </label>
+        <label class="flex flex-col gap-1">
+          <span class="text-xs text-txtsecondary">Guidance (0 = default)</span>
+          <input
+            type="number"
+            class="px-2 py-1 rounded border border-gray-200 dark:border-white/10 bg-surface focus:outline-none focus:ring-2 focus:ring-primary"
+            bind:value={$guidanceStore}
+            min="0"
+            max="30"
+            step="0.5"
+          />
+        </label>
+        <label class="flex flex-col gap-1">
+          <span class="text-xs text-txtsecondary">Seed (-1 = random)</span>
+          <input
+            type="number"
+            class="px-2 py-1 rounded border border-gray-200 dark:border-white/10 bg-surface focus:outline-none focus:ring-2 focus:ring-primary"
+            bind:value={$seedStore}
+            min="-1"
+          />
+        </label>
+      </div>
+
+      <label class="flex flex-col gap-1">
+        <span class="text-xs text-txtsecondary">Negative Prompt</span>
+        <textarea
+          class="px-2 py-1 rounded border border-gray-200 dark:border-white/10 bg-surface focus:outline-none focus:ring-2 focus:ring-primary resize-y text-sm"
+          bind:value={$negativePromptStore}
+          rows="2"
+          placeholder="Elements to avoid..."
+        ></textarea>
+      </label>
+    </div>
+  {/if}
+
+  <!-- Empty state for no models configured -->
+  {#if !hasModels}
+    <div class="flex-1 flex items-center justify-center text-txtsecondary">
+      <p>No models configured. Add models to your configuration to generate videos.</p>
+    </div>
+  {:else}
+    <!-- Video display area -->
+    <div
+      class="flex-1 overflow-auto mb-4 flex items-center justify-center bg-surface border border-gray-200 dark:border-white/10 rounded"
+    >
+      {#if isGenerating}
+        <div class="text-center text-txtsecondary">
+          <div
+            class="inline-block w-8 h-8 border-4 border-primary border-t-transparent rounded-full animate-spin mb-2"
+          ></div>
+          <p>Generating video… {formatElapsed(elapsedSeconds)}</p>
+          <p class="text-xs mt-1">
+            Video generation takes minutes — a cold model load takes longer still.
+          </p>
+        </div>
+      {:else if error}
+        <div class="text-center text-red-500 p-4">
+          <p class="font-medium">Error</p>
+          <p class="text-sm mt-1 break-all">{error}</p>
+        </div>
+      {:else if videoUrl}
+        <div class="relative max-w-full max-h-full flex items-center justify-center">
+          <!-- svelte-ignore a11y_media_has_caption -->
+          <video src={videoUrl} class="max-w-full max-h-full object-contain" controls autoplay loop></video>
+          <button
+            class="absolute bottom-2 right-2 p-2 bg-black/60 hover:bg-black/80 text-white rounded-full transition-colors"
+            onclick={downloadVideo}
+            aria-label="Download video"
+            title="Download ({formatBytes(videoBytes)})"
+          >
+            <svg class="w-5 h-5" fill="none" stroke="currentColor" viewBox="0 0 24 24">
+              <path
+                stroke-linecap="round"
+                stroke-linejoin="round"
+                stroke-width="2"
+                d="M4 16v1a3 3 0 003 3h10a3 3 0 003-3v-1m-4-4l-4 4m0 0l-4-4m4 4V4"
+              ></path>
+            </svg>
+          </button>
+        </div>
+      {:else}
+        <div class="text-center text-txtsecondary">
+          <p>Enter a prompt below to generate a short video clip</p>
+          <p class="text-xs mt-1">Attach an image to animate it (image-to-video)</p>
+        </div>
+      {/if}
+    </div>
+
+    <!-- Init image chip -->
+    {#if initImagePreview}
+      <div class="shrink-0 flex items-center gap-2 mb-2">
+        <img src={initImagePreview} alt="Conditioning frame" class="h-12 rounded border border-gray-200 dark:border-white/10" />
+        <span class="text-xs text-txtsecondary">Animating this image (image-to-video)</span>
+        <button
+          class="px-1.5 py-0.5 text-xs rounded border border-gray-200 dark:border-white/10 hover:bg-red-500 hover:text-white hover:border-red-500 transition-colors"
+          onclick={clearInitImage}
+          disabled={isGenerating}
+        >
+          Remove
+        </button>
+      </div>
+    {/if}
+
+    <!-- Prompt input area -->
+    <div class="shrink-0 flex flex-col md:flex-row gap-2">
+      <ExpandableTextarea
+        bind:value={prompt}
+        placeholder="Describe the video you want to generate..."
+        rows={3}
+        onkeydown={handleKeyDown}
+        disabled={isGenerating || !$selectedModelStore}
+      />
+      <div class="flex flex-row md:flex-col gap-2">
+        {#if isGenerating}
+          <button class="btn bg-red-500 hover:bg-red-600 text-white flex-1 md:flex-none" onclick={cancelGeneration}>
+            Cancel
+          </button>
+        {:else}
+          <button
+            class="btn bg-primary text-btn-primary-text hover:opacity-90 flex-1 md:flex-none"
+            onclick={generate}
+            disabled={!prompt.trim() || !$selectedModelStore}
+          >
+            Generate
+          </button>
+          <label
+            class="btn flex-1 md:flex-none text-center cursor-pointer {isGenerating ? 'opacity-50 pointer-events-none' : ''}"
+          >
+            Image…
+            <input
+              type="file"
+              accept="image/*"
+              class="hidden"
+              bind:this={fileInput}
+              onchange={onInitImageChange}
+            />
+          </label>
+          <button
+            class="btn flex-1 md:flex-none"
+            onclick={clearVideo}
+            disabled={!videoUrl && !error && !prompt.trim()}
+          >
+            Clear
+          </button>
+        {/if}
+      </div>
+    </div>
+  {/if}
+</div>
diff --git a/ui-svelte/src/lib/types.ts b/ui-svelte/src/lib/types.ts
index 907272ef..8d4238db 100644
--- a/ui-svelte/src/lib/types.ts
+++ b/ui-svelte/src/lib/types.ts
@@ -8,6 +8,8 @@ export interface ModelCapabilities {
   audio_speech?: boolean;
   image_generation?: boolean;
   image_to_image?: boolean;
+  video_generation?: boolean;
+  image_to_video?: boolean;
   function_calling?: boolean;
   reranker?: boolean;
 }
diff --git a/ui-svelte/src/lib/videoApi.ts b/ui-svelte/src/lib/videoApi.ts
new file mode 100644
index 00000000..971c674e
--- /dev/null
+++ b/ui-svelte/src/lib/videoApi.ts
@@ -0,0 +1,57 @@
+export interface VideoGenerationOptions {
+  negativePrompt?: string;
+  size?: string; // "WxH"
+  numFrames?: number;
+  fps?: number;
+  steps?: number;
+  guidanceScale?: number;
+  seed?: number; // -1 = random (omitted)
+  initImage?: File | null; // image-to-video conditioning frame
+}
+
+// POST /v1/videos/sync (multipart; llama-swap routes by the `model` field).
+// The response body IS the encoded clip — returns an object URL for <video>.
+// Callers must URL.revokeObjectURL() the result when done with it.
+export async function generateVideoSync(
+  model: string,
+  prompt: string,
+  options: VideoGenerationOptions = {},
+  signal?: AbortSignal
+): Promise<{ url: string; mime: string; sizeBytes: number }> {
+  const form = new FormData();
+  form.set("model", model);
+  form.set("prompt", prompt);
+  if (options.negativePrompt) form.set("negative_prompt", options.negativePrompt);
+  if (options.size) {
+    const [w, h] = options.size.split("x").map(Number);
+    if (w > 0 && h > 0) {
+      // Both conventions: vLLM-Omni reads width/height, OpenAI-shaped
+      // upstreams read size; they can never disagree.
+      form.set("width", String(w));
+      form.set("height", String(h));
+      form.set("size", options.size);
+    }
+  }
+  if (options.numFrames && options.numFrames > 0) form.set("num_frames", String(options.numFrames));
+  if (options.fps && options.fps > 0) form.set("fps", String(options.fps));
+  if (options.steps && options.steps > 0) form.set("num_inference_steps", String(options.steps));
+  if (options.guidanceScale && options.guidanceScale > 0)
+    form.set("guidance_scale", String(options.guidanceScale));
+  if (options.seed !== undefined && options.seed >= 0) form.set("seed", String(options.seed));
+  if (options.initImage) form.set("input_reference", options.initImage, options.initImage.name);
+
+  const response = await fetch("/v1/videos/sync", {
+    method: "POST",
+    body: form,
+    signal,
+  });
+
+  if (!response.ok) {
+    const errorText = await response.text();
+    throw new Error(`Video API error: ${response.status} - ${errorText}`);
+  }
+
+  const blob = await response.blob();
+  const mime = response.headers.get("Content-Type") || "video/mp4";
+  return { url: URL.createObjectURL(blob), mime, sizeBytes: blob.size };
+}
diff --git a/ui-svelte/src/routes/Playground.svelte b/ui-svelte/src/routes/Playground.svelte
index c3375311..48036143 100644
--- a/ui-svelte/src/routes/Playground.svelte
+++ b/ui-svelte/src/routes/Playground.svelte
@@ -2,12 +2,13 @@
   import { persistentStore } from "../stores/persistent";
   import ChatInterface from "../components/playground/ChatInterface.svelte";
   import ImageInterface from "../components/playground/ImageInterface.svelte";
+  import VideoInterface from "../components/playground/VideoInterface.svelte";
   import AudioInterface from "../components/playground/AudioInterface.svelte";
   import SpeechInterface from "../components/playground/SpeechInterface.svelte";
   import RerankInterface from "../components/playground/RerankInterface.svelte";
   import ConcurrencyInterface from "../components/playground/ConcurrencyInterface.svelte";
 
-  type Tab = "chat" | "images" | "speech" | "audio" | "rerank" | "concurrency";
+  type Tab = "chat" | "images" | "video" | "speech" | "audio" | "rerank" | "concurrency";
 
   const selectedTabStore = persistentStore<Tab>("playground-selected-tab", "chat");
   let mobileMenuOpen = $state(false);
@@ -15,6 +16,7 @@
   const tabs: { id: Tab; label: string }[] = [
     { id: "chat", label: "Chat" },
     { id: "images", label: "Images" },
+    { id: "video", label: "Video" },
     { id: "speech", label: "Speech" },
     { id: "audio", label: "Transcription" },
     { id: "rerank", label: "Rerank" },
@@ -92,6 +94,9 @@
     <div class="h-full" class:tab-hidden={$selectedTabStore !== "images"}>
       <ImageInterface />
     </div>
+    <div class="h-full" class:tab-hidden={$selectedTabStore !== "video"}>
+      <VideoInterface />
+    </div>
     <div class="h-full" class:tab-hidden={$selectedTabStore !== "speech"}>
       <SpeechInterface />
     </div>
diff --git a/ui-svelte/src/stores/playgroundActivity.ts b/ui-svelte/src/stores/playgroundActivity.ts
index ad067684..7a0407fd 100644
--- a/ui-svelte/src/stores/playgroundActivity.ts
+++ b/ui-svelte/src/stores/playgroundActivity.ts
@@ -2,18 +2,20 @@ import { writable, derived } from "svelte/store";
 
 const chatStreaming = writable(false);
 const imageGenerating = writable(false);
+const videoGenerating = writable(false);
 const speechGenerating = writable(false);
 const audioTranscribing = writable(false);
 const rerankLoading = writable(false);
 
 export const playgroundActivity = derived(
-  [chatStreaming, imageGenerating, speechGenerating, audioTranscribing, rerankLoading],
-  ([$chat, $image, $speech, $audio, $rerank]) => $chat || $image || $speech || $audio || $rerank
+  [chatStreaming, imageGenerating, videoGenerating, speechGenerating, audioTranscribing, rerankLoading],
+  ([$chat, $image, $video, $speech, $audio, $rerank]) => $chat || $image || $video || $speech || $audio || $rerank
 );
 
 export const playgroundStores = {
   chatStreaming,
   imageGenerating,
+  videoGenerating,
   speechGenerating,
   audioTranscribing,
   rerankLoading,