Merge pull request #149 from pmbstyle/codex/add-deepseek-provider

Add DeepSeek provider

Author: pmbstyle

.env-example

@@ -1,6 +1,9 @@
 #OpenAI API Key
 VITE_OPENAI_API_KEY = ''
 
+#DeepSeek API Key
+VITE_DEEPSEEK_API_KEY = ''
+
 #Groq whisper API Key
 VITE_GROQ_API_KEY = ''
 

README.md

@@ -84,7 +84,7 @@ With the local STT model, you can set a **wake-up word** (like "Hey, Siri").
 
 Fully customizable settings interface:
 
-* LLM provider selection between OpenAI, OpenRouter, Z.ai(coding plan), Minimax(token plan), Ollama, LM Studio
+* LLM provider selection between OpenAI, OpenRouter, DeepSeek, Z.ai(coding plan), Minimax(token plan), Ollama, LM Studio
 * Cloud or local TTS, STT, Embeddings
 * Model choice & parameters (temperature, top\_p, history, etc)
 * Prompt and summarization tuning
@@ -131,7 +131,7 @@ Follow the [Setup Instructions](https://github.com/pmbstyle/Alice/blob/main/docs
 * **Frontend:** [Vue.js](https://vuejs.org/), [TailwindCSS](https://tailwindcss.com/)
 * **Desktop Shell:** [Electron](https://www.electronjs.org/)
 * **State Management:** [Pinia](https://pinia.vuejs.org/)
-* **AI APIs:** [OpenAI](https://platform.openai.com/), [OpenRouter](https://openrouter.ai/), [Groq](https://console.groq.com/)
+* **AI APIs:** [OpenAI](https://platform.openai.com/), [OpenRouter](https://openrouter.ai/), [DeepSeek](https://platform.deepseek.com/), [Groq](https://console.groq.com/)
 * **Backend:** [Go](https://go.dev/)
 * **Vector search engine**: [hnswlib-node](https://github.com/nmslib/hnswlib)
 * **Local storage**: [better-sqlite3](https://github.com/WiseLibs/better-sqlite3)

docs/setupInstructions.md

@@ -14,7 +14,7 @@ Run in your terminal `xattr -cr "/Applications/Alice AI App.app"`
 
 # AI Provider Setup
 
-Alice supports OpenAI, OpenRouter and local LLM inference.
+Alice supports OpenAI, OpenRouter, DeepSeek, MiniMax, Z.ai, and local LLM inference.
 
 ## OpenAI (Default)
 
@@ -31,6 +31,14 @@ Alice supports OpenAI, OpenRouter and local LLM inference.
 - Models automatically include web search capabilities
 - No image generation support (use OpenAI provider for image-gen)
 
+## DeepSeek (Alternative)
+
+- Go to [DeepSeek API Platform](https://platform.deepseek.com/api_keys) to get your DeepSeek API key
+- Select "AI Provider" in Core Settings as DeepSeek
+- Use the default base URL `https://api.deepseek.com`
+- DeepSeek powers chat inference only; cloud TTS/STT/embeddings still require OpenAI or local voice and memory mode
+- No image generation support (use OpenAI provider for image-gen)
+
 ## Local Ollama / LM studio
 
 - Run Ollama or LM studio

electron/main/settingsManager.ts

@@ -13,6 +13,7 @@ export interface AppSettings {
   VITE_OPENROUTER_API_KEY?: string
   VITE_ZAI_API_KEY?: string
   VITE_MINIMAX_API_KEY?: string
+  VITE_DEEPSEEK_API_KEY?: string
   VITE_GROQ_API_KEY?: string
   sttProvider?: 'openai' | 'groq' | 'transformers'
   aiProvider?:
@@ -22,6 +23,7 @@ export interface AppSettings {
     | 'lm-studio'
     | 'zai'
     | 'minimax'
+    | 'deepseek'
 
   // Transformers STT settings
   transformersModel?: string
@@ -35,6 +37,7 @@ export interface AppSettings {
   lmStudioBaseUrl?: string
   zaiBaseUrl?: string
   minimaxBaseUrl?: string
+  deepseekBaseUrl?: string
 
   assistantModel?: string
   assistantSystemPrompt?: string

src/components/settings/AssistantSettingsTab.vue

@@ -68,6 +68,9 @@
                 (currentSettings.aiProvider === 'minimax' &&
                   currentSettings.VITE_MINIMAX_API_KEY &&
                   currentSettings.minimaxBaseUrl) ||
+                (currentSettings.aiProvider === 'deepseek' &&
+                  currentSettings.VITE_DEEPSEEK_API_KEY &&
+                  currentSettings.deepseekBaseUrl) ||
                 (currentSettings.aiProvider === 'ollama' &&
                   currentSettings.ollamaBaseUrl) ||
                 (currentSettings.aiProvider === 'lm-studio' &&
@@ -306,6 +309,9 @@
                 (currentSettings.aiProvider === 'minimax' &&
                   currentSettings.VITE_MINIMAX_API_KEY &&
                   currentSettings.minimaxBaseUrl) ||
+                (currentSettings.aiProvider === 'deepseek' &&
+                  currentSettings.VITE_DEEPSEEK_API_KEY &&
+                  currentSettings.deepseekBaseUrl) ||
                 (currentSettings.aiProvider === 'ollama' &&
                   currentSettings.ollamaBaseUrl) ||
                 (currentSettings.aiProvider === 'lm-studio' &&
@@ -519,6 +525,7 @@ const getProviderDisplayName = (provider: string): string => {
     'lm-studio': 'LM Studio',
     zai: 'Z.ai',
     minimax: 'MiniMax',
+    deepseek: 'DeepSeek',
   }
   return providerNames[provider] || provider
 }

src/components/settings/CoreSettingsTab.vue

@@ -21,6 +21,7 @@
             <option value="openrouter">OpenRouter</option>
             <option value="zai">Z.ai (Coding Plan)</option>
             <option value="minimax">MiniMax</option>
+            <option value="deepseek">DeepSeek</option>
             <option value="ollama">Ollama (Local)</option>
             <option value="lm-studio">LM Studio (Local)</option>
           </select>
@@ -197,6 +198,37 @@
             OpenAI-compatible endpoint for MiniMax token/coding plans.
           </p>
         </div>
+        <div v-if="currentSettings.aiProvider === 'deepseek'">
+          <label for="deepseek-key" class="block mb-1 text-sm"
+            >DeepSeek API Key *</label
+          >
+          <input
+            id="deepseek-key"
+            type="password"
+            v-model="currentSettings.VITE_DEEPSEEK_API_KEY"
+            class="input focus:outline-none w-full"
+            autocomplete="new-password"
+            placeholder="sk-..."
+          />
+          <p class="text-xs text-gray-400 mt-1">
+            Required for DeepSeek chat models.
+          </p>
+        </div>
+        <div v-if="currentSettings.aiProvider === 'deepseek'">
+          <label for="deepseek-url" class="block mb-1 text-sm"
+            >DeepSeek Base URL *</label
+          >
+          <input
+            id="deepseek-url"
+            type="text"
+            v-model="currentSettings.deepseekBaseUrl"
+            class="input focus:outline-none w-full"
+            placeholder="https://api.deepseek.com"
+          />
+          <p class="text-xs text-gray-400 mt-1">
+            OpenAI-compatible endpoint for DeepSeek chat completions.
+          </p>
+        </div>
         <div v-if="currentSettings.aiProvider === 'lm-studio'">
           <label for="lmstudio-url" class="block mb-1 text-sm"
             >LM Studio Base URL *</label

src/components/wizard/OnboardingWizard.vue

@@ -24,6 +24,7 @@
           @test-openrouter="testOpenRouterKey"
           @test-zai="testZAIKey"
           @test-minimax="testMiniMaxKey"
+          @test-deepseek="testDeepSeekKey"
           @test-ollama="testOllamaConnection"
           @test-lmstudio="testLMStudioConnection"
           @reset-tests="resetTestResults"
@@ -66,11 +67,13 @@ import VoiceModelsStep from './steps/VoiceModelsStep.vue'
 import FinalSetupStep from './steps/FinalSetupStep.vue'
 import OpenAI from 'openai'
 import {
+  DEEPSEEK_OPENAI_BASE_URL,
   MINIMAX_OPENAI_BASE_URL,
   PROVIDER_CONFIGS,
   ZAI_CODING_BASE_URL,
   type AIProviderKey,
 } from '../../services/llmProviders/providerCatalog'
+import { listDeepSeekModelsForConfig } from '../../services/llmProviders/deepseek'
 import { listMiniMaxModelsForConfig } from '../../services/llmProviders/minimax'
 import { listOpenAIModelsForConfig } from '../../services/llmProviders/openai'
 import { listOpenRouterModelsForConfig } from '../../services/llmProviders/openrouter'
@@ -111,6 +114,7 @@ const formData = reactive({
   VITE_OPENROUTER_API_KEY: '',
   VITE_ZAI_API_KEY: '',
   VITE_MINIMAX_API_KEY: '',
+  VITE_DEEPSEEK_API_KEY: '',
   aiProvider: 'openai' as AIProviderKey,
   assistantModel: openaiDefaults.assistantModel as string,
   summarizationModel: openaiDefaults.summarizationModel as string,
@@ -123,6 +127,7 @@ const formData = reactive({
   lmStudioBaseUrl: 'http://localhost:1234',
   zaiBaseUrl: ZAI_CODING_BASE_URL,
   minimaxBaseUrl: MINIMAX_OPENAI_BASE_URL,
+  deepseekBaseUrl: DEEPSEEK_OPENAI_BASE_URL,
   useLocalModels: false,
   availableModels: [] as string[],
   localSttLanguage: 'auto',
@@ -133,6 +138,7 @@ const isTesting = reactive({
   openrouter: false,
   zai: false,
   minimax: false,
+  deepseek: false,
   ollama: false,
   lmStudio: false,
 })
@@ -142,6 +148,7 @@ const testResult = reactive({
   openrouter: { success: false, error: '' },
   zai: { success: false, error: '' },
   minimax: { success: false, error: '' },
+  deepseek: { success: false, error: '' },
   ollama: { success: false, error: '' },
   lmStudio: { success: false, error: '' },
 })
@@ -173,7 +180,8 @@ const canContinue = computed(() => {
           formData.aiProvider === 'lm-studio' ||
           formData.aiProvider === 'openrouter' ||
           formData.aiProvider === 'zai' ||
-          formData.aiProvider === 'minimax') &&
+          formData.aiProvider === 'minimax' ||
+          formData.aiProvider === 'deepseek') &&
         !formData.VITE_OPENAI_API_KEY.trim()
       ) {
         return false
@@ -245,6 +253,8 @@ const fetchAvailableModels = async () => {
       baseURL = formData.zaiBaseUrl
     } else if (formData.aiProvider === 'minimax') {
       baseURL = formData.minimaxBaseUrl
+    } else if (formData.aiProvider === 'deepseek') {
+      baseURL = formData.deepseekBaseUrl
     } else {
       return
     }
@@ -263,6 +273,20 @@ const fetchAvailableModels = async () => {
       return
     }
 
+    if (formData.aiProvider === 'deepseek') {
+      const models = await listDeepSeekModelsForConfig(
+        formData.VITE_DEEPSEEK_API_KEY,
+        baseURL
+      )
+      formData.availableModels = models.map(model => model.id)
+
+      if (formData.availableModels.length > 0) {
+        formData.assistantModel = formData.availableModels[0]
+        formData.summarizationModel = formData.availableModels[0]
+      }
+      return
+    }
+
     if (formData.aiProvider === 'zai') {
       const models = await listZAIModelsForConfig(
         formData.VITE_ZAI_API_KEY,
@@ -414,6 +438,39 @@ const testMiniMaxKey = async () => {
   }
 }
 
+const testDeepSeekKey = async () => {
+  if (!formData.VITE_DEEPSEEK_API_KEY.trim()) {
+    testResult.deepseek.error = 'API Key cannot be empty.'
+    testResult.deepseek.success = false
+    return
+  }
+  if (!formData.deepseekBaseUrl.trim()) {
+    testResult.deepseek.error = 'Base URL cannot be empty.'
+    testResult.deepseek.success = false
+    return
+  }
+
+  isTesting.deepseek = true
+  testResult.deepseek.error = ''
+  testResult.deepseek.success = false
+
+  try {
+    await fetchAvailableModels()
+    testResult.deepseek.success = true
+  } catch (e: any) {
+    testResult.deepseek.error =
+      'API key or OpenAI-compatible endpoint is invalid or has no permissions.'
+    if (e.message?.includes('401')) {
+      testResult.deepseek.error = 'Invalid API key - please check your key.'
+    } else if (e.message?.includes('429')) {
+      testResult.deepseek.error =
+        'Rate limit exceeded - please try again later.'
+    }
+  } finally {
+    isTesting.deepseek = false
+  }
+}
+
 const testOllamaConnection = async () => {
   if (!formData.ollamaBaseUrl.trim()) {
     testResult.ollama.error = 'Ollama Base URL cannot be empty.'
@@ -499,6 +556,8 @@ const resetTestResults = () => {
   testResult.zai.error = ''
   testResult.minimax.success = false
   testResult.minimax.error = ''
+  testResult.deepseek.success = false
+  testResult.deepseek.error = ''
   testResult.ollama.success = false
   testResult.ollama.error = ''
   testResult.lmStudio.success = false
@@ -522,6 +581,12 @@ const isCurrentProviderTested = () => {
       Boolean(formData.assistantModel) &&
       Boolean(formData.summarizationModel)
     )
+  } else if (formData.aiProvider === 'deepseek') {
+    return (
+      testResult.deepseek.success &&
+      Boolean(formData.assistantModel) &&
+      Boolean(formData.summarizationModel)
+    )
   } else if (formData.aiProvider === 'ollama') {
     return (
       testResult.ollama.success &&