feat: add lemonade to provider

.env.example

@@ -79,6 +79,10 @@ XIAOMI_MODELS=
 # OLLAMA_BASE_URL=http://localhost:11434/v1
 # OLLAMA_MODELS=llama3.3,llama3.2,qwen2.5,mistral,gemma3
 
+# Lemonade local server (OpenAI-compatible, no API key required)
+# LEMONADE_BASE_URL=http://localhost:13305/v1
+# LEMONADE_MODELS=Qwen3-0.6B-GGUF,Llama-3.2-1B-Instruct-Hybrid,Qwen2.5-VL-7B-Instruct
+
 # --- TTS (Text-to-Speech) ----------------------------------------------------
 
 TTS_OPENAI_API_KEY=
@@ -99,6 +103,9 @@ TTS_MINIMAX_BASE_URL=https://api.minimaxi.com
 TTS_ELEVENLABS_API_KEY=
 TTS_ELEVENLABS_BASE_URL=
 
+# Lemonade TTS (local, no API key required)
+# TTS_LEMONADE_BASE_URL=http://localhost:13305/v1
+
 # --- ASR (Automatic Speech Recognition) --------------------------------------
 
 ASR_OPENAI_API_KEY=
@@ -107,6 +114,9 @@ ASR_OPENAI_BASE_URL=
 ASR_QWEN_API_KEY=
 ASR_QWEN_BASE_URL=
 
+# Lemonade ASR (local, WAV input only, no API key required)
+# ASR_LEMONADE_BASE_URL=http://localhost:13305/v1
+
 # --- PDF Processing -----------------------------------------------------------
 
 PDF_UNPDF_API_KEY=
@@ -136,6 +146,9 @@ IMAGE_MINIMAX_BASE_URL=https://api.minimaxi.com
 IMAGE_GROK_API_KEY=
 IMAGE_GROK_BASE_URL=
 
+# Lemonade image generation (local, no API key required)
+# IMAGE_LEMONADE_BASE_URL=http://localhost:13305/v1
+
 # --- Video Generation ---------------------------------------------------------
 
 VIDEO_SEEDANCE_API_KEY=

app/api/generate/image/route.ts

@@ -16,7 +16,11 @@
  */
 
 import { NextRequest } from 'next/server';
-import { generateImage, aspectRatioToDimensions } from '@/lib/media/image-providers';
+import {
+  generateImage,
+  aspectRatioToDimensions,
+  IMAGE_PROVIDERS,
+} from '@/lib/media/image-providers';
 import { resolveImageApiKey, resolveImageBaseUrl } from '@/lib/server/provider-config';
 import type { ImageProviderId, ImageGenerationOptions } from '@/lib/media/types';
 import { createLogger } from '@/lib/logger';
@@ -50,7 +54,8 @@ export async function POST(request: NextRequest) {
     const apiKey = clientBaseUrl
       ? clientApiKey || ''
       : resolveImageApiKey(providerId, clientApiKey);
-    if (!apiKey) {
+    const provider = IMAGE_PROVIDERS[providerId];
+    if (provider?.requiresApiKey && !apiKey) {
       return apiError(
         'MISSING_API_KEY',
         401,

app/api/transcription/route.ts

@@ -50,12 +50,8 @@ export async function POST(req: NextRequest) {
         : resolveASRBaseUrl(effectiveProviderId, baseUrl || undefined),
     };
 
-    // Convert audio file to buffer
-    const arrayBuffer = await audioFile.arrayBuffer();
-    const buffer = Buffer.from(arrayBuffer);
-
     // Transcribe using the provider system
-    const result = await transcribeAudio(config, buffer);
+    const result = await transcribeAudio(config, audioFile);
 
     return apiSuccess({ text: result.text });
   } catch (error) {

app/api/verify-image-provider/route.ts

@@ -15,7 +15,7 @@
  */
 
 import { NextRequest } from 'next/server';
-import { testImageConnectivity } from '@/lib/media/image-providers';
+import { IMAGE_PROVIDERS, testImageConnectivity } from '@/lib/media/image-providers';
 import { resolveImageApiKey, resolveImageBaseUrl } from '@/lib/server/provider-config';
 import type { ImageProviderId } from '@/lib/media/types';
 import { apiError, apiSuccess } from '@/lib/server/api-response';
@@ -43,7 +43,8 @@ export async function POST(request: NextRequest) {
       : resolveImageApiKey(providerId, clientApiKey);
     const baseUrl = clientBaseUrl ? clientBaseUrl : resolveImageBaseUrl(providerId, clientBaseUrl);
 
-    if (!apiKey) {
+    const provider = IMAGE_PROVIDERS[providerId];
+    if (provider?.requiresApiKey && !apiKey) {
       return apiError('MISSING_API_KEY', 400, 'No API key configured');
     }
 

components/settings/asr-settings.tsx

@@ -30,6 +30,7 @@ import { Mic, MicOff, CheckCircle2, XCircle, Eye, EyeOff, Plus, Loader2 } from '
 import { cn } from '@/lib/utils';
 import { toast } from 'sonner';
 import { createLogger } from '@/lib/logger';
+import { normalizeASRUploadAudio } from '@/lib/audio/wav-utils';
 
 const log = createLogger('ASRSettings');
 
@@ -52,6 +53,7 @@ export function ASRSettings({ selectedProviderId }: ASRSettingsProps) {
   const requiresApiKey = isCustom
     ? !!providerConfig?.requiresApiKey
     : !!asrProvider?.requiresApiKey;
+  const isKeylessLocalProvider = !isCustom && !requiresApiKey && !!asrProvider?.defaultBaseUrl;
 
   const [showApiKey, setShowApiKey] = useState(false);
   const [showDeleteConfirm, setShowDeleteConfirm] = useState(false);
@@ -129,8 +131,9 @@ export function ASRSettings({ selectedProviderId }: ASRSettingsProps) {
             stream.getTracks().forEach((track) => track.stop());
             setIsProcessing(true);
             const audioBlob = new Blob(audioChunks, { type: 'audio/webm' });
+            const uploadAudio = await normalizeASRUploadAudio(selectedProviderId, audioBlob);
             const formData = new FormData();
-            formData.append('audio', audioBlob, 'recording.webm');
+            formData.append('audio', uploadAudio.blob, uploadAudio.fileName);
             formData.append('providerId', selectedProviderId);
             formData.append(
               'modelId',
@@ -207,7 +210,7 @@ export function ASRSettings({ selectedProviderId }: ASRSettingsProps) {
       )}
 
       {/* API Key & Base URL */}
-      {(requiresApiKey || isServerConfigured || isCustom) && (
+      {(requiresApiKey || isServerConfigured || isCustom || isKeylessLocalProvider) && (
         <>
           <div className="grid grid-cols-2 gap-4">
             <div className="space-y-2">
@@ -276,6 +279,7 @@ export function ASRSettings({ selectedProviderId }: ASRSettingsProps) {
             } else {
               switch (selectedProviderId) {
                 case 'openai-whisper':
+                case 'lemonade-asr':
                   endpointPath = '/audio/transcriptions';
                   break;
                 case 'qwen-asr':

components/settings/audio-settings.tsx

@@ -28,6 +28,7 @@ import azureVoicesData from '@/lib/audio/azure.json';
 import { createLogger } from '@/lib/logger';
 import { getVoxCPMVoiceOptions, useVoxCPMVoiceProfiles } from '@/lib/audio/voxcpm-voices';
 import { normalizeVoxCPMBackend, voxCPMBackendSupportsReferenceAudio } from '@/lib/audio/voxcpm';
+import { normalizeASRUploadAudio } from '@/lib/audio/wav-utils';
 
 const log = createLogger('AudioSettings');
 
@@ -44,6 +45,7 @@ function getTTSProviderName(providerId: TTSProviderId, t: (key: string) => strin
     'doubao-tts': t('settings.providerDoubaoTTS'),
     'elevenlabs-tts': t('settings.providerElevenLabsTTS'),
     'minimax-tts': t('settings.providerMiniMaxTTS'),
+    'lemonade-tts': t('settings.providerLemonadeTTS'),
     'browser-native-tts': t('settings.providerBrowserNativeTTS'),
   };
   return names[providerId];
@@ -54,6 +56,7 @@ function getASRProviderName(providerId: ASRProviderId, t: (key: string) => strin
     'openai-whisper': t('settings.providerOpenAIWhisper'),
     'browser-native': t('settings.providerBrowserNative'),
     'qwen-asr': t('settings.providerQwenASR'),
+    'lemonade-asr': t('settings.providerLemonadeASR'),
   };
   return names[providerId];
 }
@@ -330,8 +333,9 @@ export function AudioSettings({ onSave }: AudioSettingsProps = {}) {
             stream.getTracks().forEach((track) => track.stop());
 
             const audioBlob = new Blob(audioChunks, { type: 'audio/webm' });
+            const uploadAudio = await normalizeASRUploadAudio(asrProviderId, audioBlob);
             const formData = new FormData();
-            formData.append('audio', audioBlob, 'recording.webm');
+            formData.append('audio', uploadAudio.blob, uploadAudio.fileName);
             formData.append('providerId', asrProviderId);
             formData.append('language', asrLanguage);
 

components/settings/image-settings.tsx

@@ -60,6 +60,7 @@ export function ImageSettings({ selectedProviderId }: ImageSettingsProps) {
     [currentConfig?.customModels],
   );
   const isServerConfigured = !!currentConfig?.isServerConfigured;
+  const requiresApiKey = currentProvider?.requiresApiKey ?? true;
 
   const handleApiKeyChange = (apiKey: string) => {
     setImageProviderConfig(selectedProviderId, { apiKey });
@@ -179,7 +180,9 @@ export function ImageSettings({ selectedProviderId }: ImageSettingsProps) {
             variant="outline"
             size="sm"
             onClick={handleTest}
-            disabled={testLoading || (!currentConfig?.apiKey && !isServerConfigured)}
+            disabled={
+              testLoading || (requiresApiKey && !currentConfig?.apiKey && !isServerConfigured)
+            }
             className="gap-1.5"
           >
             {testLoading ? (

components/settings/index.tsx

@@ -145,6 +145,7 @@ function getTTSProviderName(providerId: TTSProviderId, t: (key: string) => strin
     'doubao-tts': t('settings.providerDoubaoTTS'),
     'elevenlabs-tts': t('settings.providerElevenLabsTTS'),
     'minimax-tts': t('settings.providerMiniMaxTTS'),
+    'lemonade-tts': t('settings.providerLemonadeTTS'),
     'browser-native-tts': t('settings.providerBrowserNativeTTS'),
   };
   return names[providerId] || providerId;
@@ -159,6 +160,7 @@ function getASRProviderName(providerId: ASRProviderId, t: (key: string) => strin
     'openai-whisper': t('settings.providerOpenAIWhisper'),
     'browser-native': t('settings.providerBrowserNative'),
     'qwen-asr': t('settings.providerQwenASR'),
+    'lemonade-asr': t('settings.providerLemonadeASR'),
   };
   return names[providerId] || providerId;
 }
@@ -171,6 +173,7 @@ const IMAGE_PROVIDER_NAMES: Record<ImageProviderId, string> = {
   'nano-banana': 'providerNanoBanana',
   'minimax-image': 'providerMiniMaxImage',
   'grok-image': 'providerGrokImage',
+  lemonade: 'providerLemonadeImage',
 };
 
 const IMAGE_PROVIDER_ICONS: Record<ImageProviderId, string> = {
@@ -180,6 +183,7 @@ const IMAGE_PROVIDER_ICONS: Record<ImageProviderId, string> = {
   'nano-banana': '/logos/gemini.svg',
   'minimax-image': '/logos/minimax.svg',
   'grok-image': '/logos/grok.svg',
+  lemonade: '/logos/lemonade.svg',
 };
 
 const VIDEO_PROVIDER_NAMES: Record<VideoProviderId, string> = {

components/settings/model-selector.tsx

@@ -75,7 +75,7 @@ export function ModelSelector({
       ([, config]) =>
         (config.requiresApiKey
           ? config.apiKey || config.isServerConfigured
-          : config.isServerConfigured || config.baseUrl) &&
+          : config.isServerConfigured || config.baseUrl || config.defaultBaseUrl) &&
         config.models.length >= 1 &&
         (config.baseUrl || config.defaultBaseUrl || config.serverBaseUrl),
     )

components/settings/tts-settings.tsx

@@ -93,6 +93,7 @@ export function TTSSettings({ selectedProviderId }: TTSSettingsProps) {
   const requiresApiKey = isCustom
     ? !!providerConfig?.requiresApiKey
     : !!ttsProvider?.requiresApiKey;
+  const isKeylessLocalProvider = !isCustom && !requiresApiKey && !!ttsProvider?.defaultBaseUrl;
 
   // When testing a non-active provider, use that provider's default voice
   // instead of the active provider's voice (which may be incompatible).
@@ -192,6 +193,7 @@ export function TTSSettings({ selectedProviderId }: TTSSettingsProps) {
     switch (selectedProviderId) {
       case 'openai-tts':
       case 'glm-tts':
+      case 'lemonade-tts':
         return '/audio/speech';
       case 'azure-tts':
         return '/cognitiveservices/v1';
@@ -225,7 +227,7 @@ export function TTSSettings({ selectedProviderId }: TTSSettingsProps) {
       )}
 
       {/* API Key & Base URL */}
-      {(requiresApiKey || isServerConfigured || isCustom || isVoxCPM) &&
+      {(requiresApiKey || isServerConfigured || isCustom || isVoxCPM || isKeylessLocalProvider) &&
         (isVoxCPM ? (
           <div className="rounded-lg border border-border/60 bg-background px-3 py-2.5">
             <div className="flex flex-col gap-2 md:flex-row md:items-end">

lib/ai/providers.ts

@@ -6,7 +6,7 @@
  * - Anthropic Claude (native)
  * - Google Gemini (native)
  * - MiniMax (Anthropic-compatible, recommended by official)
- * - OpenAI-compatible providers (DeepSeek, Qwen, Kimi, GLM, SiliconFlow, Doubao, Tencent, Xiaomi, etc.)
+ * - OpenAI-compatible providers (DeepSeek, Qwen, Kimi, GLM, SiliconFlow, Doubao, Tencent, Xiaomi, Lemonade, etc.)
  *
  * Sources:
  * - https://platform.openai.com/docs/models
@@ -980,6 +980,32 @@ export const PROVIDERS: Record<ProviderId, ProviderConfig> = {
       },
     ],
   },
+
+  lemonade: {
+    id: 'lemonade',
+    name: 'Lemonade',
+    type: 'openai',
+    defaultBaseUrl: 'http://localhost:13305/v1',
+    requiresApiKey: false,
+    icon: '/logos/lemonade.svg',
+    models: [
+      {
+        id: 'Qwen3-0.6B-GGUF',
+        name: 'Qwen3 0.6B GGUF',
+        capabilities: { streaming: true, tools: true, vision: false },
+      },
+      {
+        id: 'Llama-3.2-1B-Instruct-Hybrid',
+        name: 'Llama 3.2 1B Instruct Hybrid',
+        capabilities: { streaming: true, tools: true, vision: false },
+      },
+      {
+        id: 'Qwen2.5-VL-7B-Instruct',
+        name: 'Qwen2.5 VL 7B Instruct',
+        capabilities: { streaming: true, tools: true, vision: true },
+      },
+    ],
+  },
 };
 
 applyModelMetadata(PROVIDERS);

lib/audio/asr-providers.ts

@@ -182,6 +182,9 @@ export async function transcribeAudio(
     case 'qwen-asr':
       return await transcribeQwenASR(config, audioBuffer);
 
+    case 'lemonade-asr':
+      return await transcribeLemonadeASR(config, audioBuffer);
+
     default:
       if (isCustomASRProvider(config.providerId)) {
         return await transcribeOpenAIWhisper(config, audioBuffer);
@@ -190,6 +193,84 @@ export async function transcribeAudio(
   }
 }
 
+/**
+ * Lemonade ASR implementation (OpenAI-compatible multipart transcription).
+ *
+ * Lemonade currently supports WAV input and JSON response format.
+ */
+async function transcribeLemonadeASR(
+  config: ASRModelConfig,
+  audioBuffer: Buffer | Blob,
+): Promise<ASRTranscriptionResult> {
+  const baseUrl = (config.baseUrl || ASR_PROVIDERS['lemonade-asr'].defaultBaseUrl || '').replace(
+    /\/$/,
+    '',
+  );
+
+  const audioBlob = await toAudioBlob(audioBuffer);
+  if (!isWavAudio(audioBlob)) {
+    throw new Error(
+      'Lemonade ASR currently supports WAV input only. Recordings should be converted to WAV before upload.',
+    );
+  }
+
+  const formData = new FormData();
+  formData.set('file', audioBlob, 'audio.wav');
+  formData.set('model', config.modelId || ASR_PROVIDERS['lemonade-asr'].defaultModelId);
+  formData.set('response_format', 'json');
+  if (config.language && config.language !== 'auto') {
+    formData.set('language', config.language);
+  }
+
+  const response = await fetch(`${baseUrl}/audio/transcriptions`, {
+    method: 'POST',
+    headers: getOptionalBearerAuthHeaders(config.apiKey),
+    body: formData,
+  });
+
+  if (!response.ok) {
+    const errorText = await response.text().catch(() => response.statusText);
+    if (errorText.includes('audio is empty') || errorText.includes('too short')) {
+      return { text: '' };
+    }
+    throw new Error(`Lemonade ASR API error: ${errorText || response.statusText}`);
+  }
+
+  const data = await response.json();
+  return { text: typeof data.text === 'string' ? data.text : '' };
+}
+
+async function toAudioBlob(audioBuffer: Buffer | Blob): Promise<Blob> {
+  if (audioBuffer instanceof Blob) {
+    return audioBuffer;
+  }
+  if (audioBuffer instanceof Buffer) {
+    const arrayBuffer = audioBuffer.buffer.slice(
+      audioBuffer.byteOffset,
+      audioBuffer.byteOffset + audioBuffer.byteLength,
+    ) as ArrayBuffer;
+    return new Blob([arrayBuffer], { type: detectWavBuffer(audioBuffer) ? 'audio/wav' : '' });
+  }
+  throw new Error('Invalid audio buffer type');
+}
+
+function isWavAudio(blob: Blob): boolean {
+  return blob.type.includes('audio/wav') || blob.type.includes('audio/x-wav');
+}
+
+function detectWavBuffer(buffer: Buffer): boolean {
+  return (
+    buffer.byteLength >= 12 &&
+    buffer.toString('ascii', 0, 4) === 'RIFF' &&
+    buffer.toString('ascii', 8, 12) === 'WAVE'
+  );
+}
+
+function getOptionalBearerAuthHeaders(apiKey?: string): Record<string, string> {
+  const key = apiKey?.trim();
+  return key ? { Authorization: `Bearer ${key}` } : {};
+}
+
 /**
  * OpenAI Whisper implementation (using Vercel AI SDK)
  */