chore: update model download recommendation

apps/desktop/src/renderer/onboarding/components/screens/ModelSetupModal.tsx

@@ -44,6 +44,12 @@ export function ModelSetupModal({
   const [installedModelName, setInstalledModelName] = useState<string>("");
   const [downloadComplete, setDownloadComplete] = useState(false);
 
+  // Get recommended local model based on hardware
+  const { data: recommendedModelId = "whisper-base" } =
+    api.onboarding.getRecommendedLocalModel.useQuery(undefined, {
+      enabled: modelType === ModelType.Local && isOpen,
+    });
+
   // tRPC mutations
   const loginMutation = api.auth.login.useMutation({
     onSuccess: () => {
@@ -84,7 +90,7 @@ export function ModelSetupModal({
   // Subscribe to download progress
   api.models.onDownloadProgress.useSubscription(undefined, {
     onData: (data) => {
-      if (data.modelId === "whisper-tiny") {
+      if (data.modelId === recommendedModelId) {
         setDownloadProgress(data.progress.progress);
         setDownloadInfo({
           downloaded: data.progress.bytesDownloaded || 0,
@@ -116,7 +122,7 @@ export function ModelSetupModal({
 
     try {
       await downloadModelMutation.mutateAsync({
-        modelId: "whisper-tiny",
+        modelId: recommendedModelId,
       });
       // Progress will be handled by subscription
     } catch (err) {
@@ -188,7 +194,7 @@ export function ModelSetupModal({
           <DialogDescription>
             {modelAlreadyInstalled || downloadComplete
               ? "Ready for private, offline transcription."
-              : "Setting up Whisper Tiny for private, offline transcription"}
+              : "Setting up local model for private, offline transcription"}
           </DialogDescription>
         </DialogHeader>
 
@@ -206,7 +212,7 @@ export function ModelSetupModal({
                     : "Download Complete"}
                 </p>
                 <p className="text-sm text-muted-foreground mt-1">
-                  Using: {installedModelName || "whisper-tiny"}
+                  Using: {installedModelName || recommendedModelId}
                 </p>
               </div>
             </div>

apps/desktop/src/services/onboarding-service.ts

@@ -431,6 +431,80 @@ export class OnboardingService extends EventEmitter {
     );
   }
 
+  /**
+   * Check for high-end hardware (RTX 50 series or M3 Pro/Max/M4)
+   */
+  private hasHighEndHardware(gpuModel: string, cpuModel: string): boolean {
+    const upperGpu = gpuModel.toUpperCase();
+    const upperCpu = cpuModel.toUpperCase();
+
+    // RTX 50 series
+    const hasRtx50 = ["RTX 5060", "RTX 5070", "RTX 5080", "RTX 5090"].some(
+      (m) => upperGpu.includes(m),
+    );
+
+    // M3 Pro/Max or M4+
+    const hasM3ProMax =
+      upperCpu.includes("M3 PRO") || upperCpu.includes("M3 MAX");
+    const hasM4Plus = ["M4", "M5", "M6"].some((chip) =>
+      upperCpu.includes(`APPLE ${chip}`),
+    );
+
+    return hasRtx50 || hasM3ProMax || hasM4Plus;
+  }
+
+  /**
+   * Check for NVIDIA RTX 20 series
+   */
+  private hasNvidia20Series(gpuModel: string): boolean {
+    if (!gpuModel) return false;
+    const upperGpu = gpuModel.toUpperCase();
+    return ["RTX 2060", "RTX 2070", "RTX 2080"].some((m) =>
+      upperGpu.includes(m),
+    );
+  }
+
+  /**
+   * Check for Apple Silicon M1
+   */
+  private hasAppleSiliconM1(cpuModel: string): boolean {
+    if (process.platform !== "darwin" || process.arch !== "arm64") return false;
+    return cpuModel.toUpperCase().includes("APPLE M1");
+  }
+
+  /**
+   * Get recommended local model ID based on hardware
+   * - High-end (RTX 50, M3 Pro/Max, M4+) → whisper-large-v3-turbo
+   * - Mid-tier (RTX 30/40, M2/M3 base) → whisper-medium
+   * - Entry (RTX 20, M1) → whisper-small
+   * - Default → whisper-base
+   */
+  getRecommendedLocalModelId(): string {
+    const systemInfo = this.telemetryService.getSystemInfo();
+    const gpuModel = systemInfo?.gpu_model || "";
+    const cpuModel = systemInfo?.cpu_model || "";
+
+    // High-end: RTX 50 series or M3 Pro/Max/M4+
+    if (this.hasHighEndHardware(gpuModel, cpuModel)) {
+      return "whisper-large-v3-turbo";
+    }
+
+    // Mid-tier: RTX 30/40 series or M2/M3 base
+    if (
+      this.hasNvidia30SeriesOrBetter(gpuModel) ||
+      this.hasAppleSiliconM2OrBetter(cpuModel)
+    ) {
+      return "whisper-medium";
+    }
+
+    // Entry: RTX 20 series or M1
+    if (this.hasNvidia20Series(gpuModel) || this.hasAppleSiliconM1(cpuModel)) {
+      return "whisper-small";
+    }
+
+    return "whisper-base";
+  }
+
   /**
    * Calculate model recommendation based on system specs
    */

apps/desktop/src/services/transcription-service.ts

@@ -157,6 +157,16 @@ export class TranscriptionService {
    */
   public async isModelAvailable(): Promise<boolean> {
     try {
+      // Check if selected model is a cloud model (doesn't need download)
+      const selectedModelId = await this.modelService.getSelectedModel();
+      if (selectedModelId) {
+        const model = AVAILABLE_MODELS.find((m) => m.id === selectedModelId);
+        if (model?.provider === "Amical Cloud") {
+          return true;
+        }
+      }
+
+      // For local models, check if any are downloaded
       const modelService = this.whisperProvider["modelService"];
       const availableModels = await modelService.getValidDownloadedModels();
       return Object.keys(availableModels).length > 0;

apps/desktop/src/trpc/routers/onboarding.ts

@@ -65,6 +65,21 @@ export const onboardingRouter = createRouter({
     },
   ),
 
+  /**
+   * Get recommended local model ID based on hardware
+   */
+  getRecommendedLocalModel: procedure.query(({ ctx }): string => {
+    const { serviceManager } = ctx;
+    if (!serviceManager) {
+      return "whisper-base";
+    }
+    const onboardingService = serviceManager.getOnboardingService();
+    if (!onboardingService) {
+      return "whisper-base";
+    }
+    return onboardingService.getRecommendedLocalModelId();
+  }),
+
   /**
    * Check if onboarding is needed
    */