fix litellm metadata model discovery

src/routes/adminConfig.js

@@ -9,9 +9,10 @@ var { authMiddleware, adminMiddleware } = require('../middleware/auth');
 var PROMPTS = require('../utils/prompts');
 var logger = require('../utils/logger');
 var { gatewayUrl } = require('../utils/errors');
-var { GOOGLE_TTS_VOICES, OPENAI_TTS_VOICES, getTTSEnvProvider, getLiteLLMTTSModels, getTTSProvider, getTTSVoiceLists } = require('../utils/ttsProvider');
+var { GOOGLE_TTS_VOICES, getTTSEnvProvider, getLiteLLMTTSModels, getTTSProvider, getTTSVoiceLists } = require('../utils/ttsProvider');
 var { getLiteLLMHeaders } = require('../utils/litellm');
 var { getSTTDependencies, getLiteLLMSTTModels, getSTTModelLists, getSTTProvider } = require('../utils/sttProvider');
+var { getLiteLLMEmbeddingModels } = require('../utils/embeddings');
 
 router.use(authMiddleware);
 
@@ -570,7 +571,16 @@ router.post('/config/tts/test', async function(req, res) {
       var db = require('../db/database');
       var adminModel = await db.getSetting('tts.model') || '';
       var adminVoice = await db.getSetting('tts.voice') || '';
-      if (!adminModel && adminVoice && (adminVoice.indexOf('/') !== -1 || /^(openai-tts|tts-1|openai-gpt.*tts)/i.test(adminVoice))) { adminModel = adminVoice; adminVoice = ''; }
+      if (!adminModel && adminVoice) {
+        try {
+          var liteLLMBase = (process.env.LITELLM_API_BASE || '').replace(/\/+$/, '').replace(/\/v1\/?$/, '');
+          var modelInfoResp = await axios.get(liteLLMBase + '/model/info', { headers: getLiteLLMHeaders(), timeout: 10000 });
+          if (getLiteLLMTTSModels(modelInfoResp.data && modelInfoResp.data.data).indexOf(adminVoice) !== -1) {
+            adminModel = adminVoice;
+            adminVoice = '';
+          }
+        } catch (e) { logger.warn('LiteLLM TTS mode lookup failed: ' + e.message); }
+      }
       var resolvedVoice = voice || adminVoice || '';
       var rawModel = adminModel || process.env.LITELLM_TTS_MODEL || 'tts-1';
       var ttsModel = rawModel;
@@ -659,7 +669,6 @@ router.get('/config/stt/discover', async function(req, res) {
     }
 
     if (provider === 'litellm' && process.env.LITELLM_API_BASE) {
-      // base URL handled by gatewayUrl() helper
       try {
         var liteLLMBase = (process.env.LITELLM_API_BASE || '').replace(/\/+$/, '').replace(/\/v1\/?$/, '');
         var sResp = await axios.get(liteLLMBase + '/model/info', { headers: getLiteLLMHeaders(), timeout: 10000 });
@@ -723,9 +732,13 @@ router.post('/config/stt/test', async function(req, res) {
       var db = require('../db/database');
       var adminSttModel = await db.getSetting('stt.model') || '';
       var sttModel = adminSttModel || process.env.LITELLM_STT_MODEL || 'gemini-2.0-flash';
-      // base URL handled by gatewayUrl() helper
 
-      var isTranscriptionModel = /whisper|parakeet|deepgram|nova|groq|scribe|elevenlabs|transcri/i.test(sttModel);
+      var isTranscriptionModel = false;
+      try {
+        var liteLLMBase = (process.env.LITELLM_API_BASE || '').replace(/\/+$/, '').replace(/\/v1\/?$/, '');
+        var modelInfoResp = await axios.get(liteLLMBase + '/model/info', { headers: getLiteLLMHeaders(), timeout: 10000 });
+        isTranscriptionModel = getLiteLLMSTTModels(modelInfoResp.data && modelInfoResp.data.data).indexOf(sttModel) !== -1;
+      } catch (e) { logger.warn('LiteLLM STT mode lookup failed: ' + e.message); }
 
       if (isTranscriptionModel) {
         var ext = mime.split('/')[1] || 'webm';
@@ -820,23 +833,20 @@ router.get('/config/embeddings/discover', async function(req, res) {
     ];
 
     if (provider === 'litellm') {
-      // base URL handled by gatewayUrl() helper
       try {
-        var eResp = await axios.get(gatewayUrl('/models'), { headers: getLiteLLMHeaders(), timeout: 10000 });
+        var liteLLMBase = (process.env.LITELLM_API_BASE || '').replace(/\/+$/, '').replace(/\/v1\/?$/, '');
-        if (eResp.data && eResp.data.data) {
+        var eResp = await axios.get(liteLLMBase + '/model/info', { headers: getLiteLLMHeaders(), timeout: 10000 });
-          eResp.data.data.forEach(function(m) {
+        getLiteLLMEmbeddingModels(eResp.data && eResp.data.data).forEach(function(m) {
-            if ((m.id || '').toLowerCase().indexOf('embed') !== -1) {
+          discovered.push({ id: m.id, name: m.name, dims: m.dims, source: 'gateway-api' });
-              var dims = m.id.indexOf('3-large') !== -1 ? 3072 : m.id.indexOf('3-small') !== -1 ? 1536 : m.id.indexOf('005') !== -1 ? 768 : '?';
+        });
-              discovered.push({ id: m.id, name: m.id, dims: dims, source: 'gateway-api' });
+      } catch(e) { logger.warn('LiteLLM embedding metadata discovery failed: ' + e.message); }
-            }
+      if (discovered.length === 0) {
-          });
+        knownVertex.forEach(function(m) {
-        }
+          if (!discovered.find(function(d) { return d.id === m.id; })) {
-      } catch(e) { logger.warn('LiteLLM embedding discovery failed: ' + e.message); }
+            discovered.push(Object.assign({ source: 'builtin' }, m));
-      knownVertex.forEach(function(m) {
+          }
-        if (!discovered.find(function(d) { return d.id === m.id; })) {
+        });
-          discovered.push(Object.assign({ source: 'builtin' }, m));
+      }
-        }
-      });
     }
 
     if (provider === 'vertex') {

src/utils/embeddings.js

@@ -239,6 +239,34 @@ function stripHtml(html) {
   return html.replace(/<[^>]*>/g, ' ').replace(/\s+/g, ' ').trim();
 }
 
+function getLiteLLMModelId(model) {
+  if (!model) return '';
+  if (typeof model === 'string') return model;
+  return model.id || model.model_name || '';
+}
+
+function isLiteLLMEmbeddingModel(model) {
+  var mode = model && model.model_info && model.model_info.mode ? String(model.model_info.mode) : '';
+  return mode === 'embedding';
+}
+
+function getLiteLLMEmbeddingDimensions(model) {
+  var info = model && model.model_info ? model.model_info : {};
+  var dims = info.output_vector_size || info.dimensions || info.embedding_dimensions || model.output_vector_size || model.dimensions || model.embedding_dimensions;
+  var parsed = parseInt(dims, 10);
+  return Number.isFinite(parsed) ? parsed : '?';
+}
+
+function getLiteLLMEmbeddingModels(models) {
+  return (models || [])
+    .filter(isLiteLLMEmbeddingModel)
+    .map(function(model) {
+      var id = getLiteLLMModelId(model);
+      return { id: id, name: id, dims: getLiteLLMEmbeddingDimensions(model) };
+    })
+    .filter(function(model) { return !!model.id; });
+}
+
 /**
  * Check if embeddings are available (provider configured)
  */
@@ -254,8 +282,10 @@ function isEmbeddingsAvailable() {
 module.exports = {
   generateEmbedding,
   generateContentEmbedding,
+  getLiteLLMEmbeddingModels,
   searchSimilar,
   isEmbeddingsAvailable,
+  isLiteLLMEmbeddingModel,
   DEFAULT_MODEL,
   DEFAULT_DIMS
 };

src/utils/ttsProvider.js

@@ -28,10 +28,6 @@ function getTTSVoiceLists() {
   };
 }
 
-function isOpenAITTSModel(modelId) {
-  return /^openai[-/](tts-1|tts-1-hd|gpt-4o-mini-tts|gpt-audio)/.test(String(modelId || '').toLowerCase());
-}
-
 function isLiteLLMTTSModel(model) {
   var mode = model && model.model_info && model.model_info.mode ? String(model.model_info.mode) : '';
   return mode === 'audio_speech';
@@ -52,6 +48,5 @@ module.exports = {
   getLiteLLMTTSModels,
   getTTSProvider,
   getTTSVoiceLists,
-  isLiteLLMTTSModel,
+  isLiteLLMTTSModel
-  isOpenAITTSModel
 };

test/embeddings-provider.test.js

@@ -0,0 +1,25 @@
+const { test } = require('node:test');
+const assert = require('node:assert/strict');
+
+test('LiteLLM embedding discovery uses embedding metadata only', () => {
+  const embeddings = require('../src/utils/embeddings');
+  assert.equal(embeddings.isLiteLLMEmbeddingModel({ id: 'openai-text-embedding-3-large' }), false);
+  assert.equal(embeddings.isLiteLLMEmbeddingModel({ id: 'mistral-codestral-embed' }), false);
+  assert.equal(embeddings.isLiteLLMEmbeddingModel({ model_name: 'openai-text-embedding-3-large', model_info: { mode: 'embedding' } }), true);
+  assert.equal(embeddings.isLiteLLMEmbeddingModel({ model_name: 'local-parakeet-v3', model_info: { mode: 'audio_transcription' } }), false);
+});
+
+test('LiteLLM embedding extraction includes metadata dimensions', () => {
+  const embeddings = require('../src/utils/embeddings');
+  assert.deepEqual(embeddings.getLiteLLMEmbeddingModels([
+    { model_name: 'openai-text-embedding-3-large', model_info: { mode: 'embedding', output_vector_size: 3072 } },
+    { model_name: 'openai-text-embedding-3-small', model_info: { mode: 'embedding', dimensions: '1536' } },
+    { model_name: 'mistral-embed', model_info: { mode: 'embedding' } },
+    { model_name: 'looks-like-embed' },
+    { model_name: 'local-kokoro-tts', model_info: { mode: 'audio_speech' } }
+  ]), [
+    { id: 'openai-text-embedding-3-large', name: 'openai-text-embedding-3-large', dims: 3072 },
+    { id: 'openai-text-embedding-3-small', name: 'openai-text-embedding-3-small', dims: 1536 },
+    { id: 'mistral-embed', name: 'mistral-embed', dims: '?' }
+  ]);
+});

test/tts-provider.test.js

@@ -48,19 +48,15 @@ test('LiteLLM TTS voice list stays OpenAI-compatible', () => {
   ]);
 });
 
-test('TTS model discovery keeps OpenAI speech models only', () => {
-  const ttsProvider = require('../src/utils/ttsProvider');
-  assert.equal(ttsProvider.isOpenAITTSModel('openai/tts-1'), true);
-  assert.equal(ttsProvider.isOpenAITTSModel('openai-tts-1-hd'), true);
-  assert.equal(ttsProvider.isOpenAITTSModel('openai-gpt-4o-mini-tts'), true);
-  assert.equal(ttsProvider.isOpenAITTSModel('mistral-voxtral-mini-tts'), false);
-  assert.equal(ttsProvider.isOpenAITTSModel('vertex-gemini-2.5-flash-tts'), false);
-});
-
 test('TTS model discovery uses audio_speech metadata only', () => {
   const ttsProvider = require('../src/utils/ttsProvider');
+  assert.equal(ttsProvider.isOpenAITTSModel, undefined);
   assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'local-chatterbox-turbo' }), false);
   assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'local-kokoro-tts' }), false);
+  assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'openai-tts-1' }), false);
+  assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'openai-gpt-4o-mini-tts' }), false);
+  assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'mistral-voxtral-mini-tts' }), false);
+  assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'vertex-gemini-2.5-flash-tts' }), false);
   assert.equal(ttsProvider.isLiteLLMTTSModel({ model_name: 'local-chatterbox-turbo', model_info: { mode: 'audio_speech' } }), true);
   assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'custom-provider-model', model_info: { mode: 'audio_speech' } }), true);
   assert.equal(ttsProvider.isLiteLLMTTSModel({ id: 'local-parakeet-v3', model_info: { mode: 'audio_transcription' } }), false);