Merge pull request #58 from donvito/feature/gemini-support

feature/gemini-support

Author: donvito

README.md

@@ -49,7 +49,7 @@ More to come...check swagger docs for updated endpoints.
 | [OpenRouter](https://openrouter.ai/) | Open source and private models |  Available |
 | [Vercel AI Gateway](https://vercel.com/ai-gateway) | Open source and private models | Available |
 | [LlamaCpp](https://github.com/ggml-org/llama.cpp) | Local models via llama.cpp server (self-hosted) | Available |
-| [Google](https://ai.google.dev/) | Gemini models | In Progress |
+| [Google Gemini](https://ai.google.dev/) | Gemini models via OpenAI-compatible interface | Available |
 
 
 ## Run the project
@@ -93,6 +93,7 @@ DEFAULT_ACCESS_TOKEN=your-secret-api-key
 OPENAI_API_KEY=your-openai-api-key
 ANTHROPIC_API_KEY=your-anthropic-api-key
 OPENROUTER_API_KEY=your-openrouter-api-key
+GOOGLE_AI_API_KEY=your-google-ai-api-key
 ```
 You need to configure at least one provider api key. Otherwise, the app will not start.
 
@@ -144,6 +145,10 @@ OPENAI_API_KEY=your-openai-api-key
 # Anthropic Configuration
 ANTHROPIC_API_KEY=your-anthropic-api-key
 
+# Google Gemini Configuration
+GOOGLE_AI_API_KEY=your-google-ai-api-key
+GEMINI_MODEL=gemini-2.5-flash-lite
+
 # Ollama Configuration
 OLLAMA_ENABLED=true
 OLLAMA_BASE_URL=http://localhost:11434
@@ -161,8 +166,27 @@ LMSTUDIO_ENABLED=true
 LMSTUDIO_BASE_URL=http://localhost:1234
 
 # You can change LMSTUDIO_BASE_URL to use a remote LM Studio instance
+
+# OpenRouter Configuration 
+OPENROUTER_API_KEY=your-openrouter-api-key
 ```
 
+### Google Gemini Setup
+
+To use Google Gemini models:
+
+1. Get your API key from [Google AI Studio](https://makersuite.google.com/app/apikey)
+2. Set `GOOGLE_AI_API_KEY` in your `.env` file
+3. Optionally configure `GEMINI_MODEL` (defaults to `gemini-2.5-flash-lite`)
+
+Available Gemini models:
+- `gemini-2.5-flash-lite` (default)
+- `gemini-2.5-flash`
+- `gemini-2.5-pro`
+- `gemini-pro-vision`
+
+**Note**: The Gemini provider uses Google's OpenAI-compatible interface to maintain compatibility with AI SDK v4.
+
 **Important:** Make sure to add `.env` to your `.gitignore` file to avoid committing sensitive information to version control.
 
 
@@ -196,7 +220,26 @@ You can access demos at http://localhost:3000/api/demos
 ## Provider and Model Selection
 You need to send the service and model name in the request body. See examples in the swagger docs.
 
-For example, to summarize text using qwen2.5-coder model with Ollama as provider, you can use the following curl command:
+For example, to summarize text using Gemini model with Google as provider, you can use the following curl command:
+
+```curl
+curl --location 'http://localhost:3000/api/v1/summarize' \
+--header 'Content-Type: application/json' \
+--header 'Accept: application/json' \
+--data '{
+    "payload": {
+        "text": "Text to summarize",
+        "maxLength": 100
+    },
+    "config": {
+        "provider": "google",
+        "model": "gemini-2.5-flash-lite",
+        "temperature": 0
+    }
+}'
+```
+
+Or with Ollama:
 
 ```curl
 curl --location 'http://localhost:3000/api/v1/summarize' \

bun.lock

@@ -136,13 +136,14 @@ export async function configureAuth(app: OpenAPIHono): Promise<void> {
 
 export async function checkLLMProvidersAvailability() {
 
-    // Check if OpenAI, Anthropic, or OpenRouter are available via API keys
+    // Check if OpenAI, Anthropic, OpenRouter, or Google are available via API keys
     const openaiApiKey = process.env.OPENAI_API_KEY;
     const anthropicApiKey = process.env.ANTHROPIC_API_KEY;
     const openrouterApiKey = process.env.OPENROUTER_API_KEY;
     const aigatewayApiKey = process.env.AI_GATEWAY_API_KEY;
+    const geminiApiKey = process.env.GOOGLE_AI_API_KEY;
 
-    if (!openaiApiKey && !anthropicApiKey && !openrouterApiKey && !aigatewayApiKey) {
+    if (!openaiApiKey && !anthropicApiKey && !openrouterApiKey && !aigatewayApiKey && !geminiApiKey) {
         throw new Error('No API keys found for external LLM providers');
     }
 
@@ -151,6 +152,7 @@ export async function checkLLMProvidersAvailability() {
     if (anthropicApiKey) availableProviders.push('Anthropic');
     if (openrouterApiKey) availableProviders.push('OpenRouter');
     if (aigatewayApiKey) availableProviders.push('AIGateway');
+    if (geminiApiKey) availableProviders.push('Google');
 
     return availableProviders;
 }

src/app.ts

@@ -42,6 +42,8 @@
         { "name": "openai/gpt-4o-mini", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "OpenRouter proxy to GPT-4o mini." },
         { "name": "google/gemini-2.0-flash-lite-001", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "Google Gemini 2.0 Flash Lite via OpenRouter with Q&A capabilities." },
         { "name": "google/gemini-2.5-flash-lite", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "Google Gemini 2.5 Flash Lite via OpenRouter with 1M token context window for large document Q&A." },
+        { "name": "google/gemini-2.5-flash", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "Google Gemini 2.5 Flash via OpenRouter with advanced capabilities." },
+        { "name": "google/gemini-2.5-pro", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "Google Gemini 2.5 Pro via OpenRouter with enhanced reasoning." },
         { "name": "openai/gpt-oss-20b", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "OpenRouter proxy to GPT OSS 20B" },
         { "name": "openai/gpt-oss-120b", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "OpenRouter proxy to GPT OSS 120B" },
         { "name": "nvidia/nemotron-nano-9b-v2", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "NVIDIA Nemotron Nano 9B v2; compact high-performance model." }
@@ -63,6 +65,14 @@
 
       ]
     },
+    "google": {
+      "enabled": true,
+      "models": [
+        { "name": "gemini-2.5-flash-lite", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "vision", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "Google Gemini 2.5 Flash Lite with fast processing and vision capabilities." },
+        { "name": "gemini-2.5-flash", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "vision", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "Google Gemini 2.5 Flash with advanced multimodal capabilities." },
+        { "name": "gemini-2.5-pro", "capabilities": ["summarize", "pdf-summarizer", "rewrite", "compose", "planning", "keywords", "sentiment", "vision", "askText", "emailReply", "translate", "meetingNotes", "outline"], "notes": "Google Gemini 2.5 Pro with enhanced reasoning and vision support." }
+      ]
+    },
     "llamacpp": {
       "enabled": true,
       "models": [

src/config/models.json

@@ -3,6 +3,7 @@ import * as path from 'path'
 
 export type ModelCapability =
   | 'summarize'
+  | 'pdf-summarizer'
   | 'rewrite'
   | 'compose'
   | 'keywords'

src/config/models.ts

@@ -51,6 +51,11 @@ export interface LlamaCppConfig extends ServiceConfig {
   timeout?: number;
 }
 
+export interface GoogleConfig extends ServiceConfig {
+  apiKey: string;
+  model: string;
+}
+
 // OpenAI Configuration
 export const openaiConfig: OpenAIConfig = {
   name: 'OpenAI',
@@ -123,8 +128,17 @@ export const llamacppConfig: LlamaCppConfig = {
   timeout: parseInt(process.env.LLAMACPP_TIMEOUT || '30000'),
 };
 
+// Google Gemini Configuration
+export const googleConfig: GoogleConfig = {
+  name: 'Google',
+  enabled: !!process.env.GOOGLE_AI_API_KEY,
+  priority: 8,
+  apiKey: process.env.GOOGLE_AI_API_KEY || '',
+  model: process.env.GEMINI_MODEL || 'gemini-2.5-flash-lite',
+};
+
 // Available services
-export const availableServices = [openaiConfig, anthropicConfig, ollamaConfig, openrouterConfig, lmstudioConfig, aigatewayConfig, llamacppConfig];
+export const availableServices = [openaiConfig, anthropicConfig, ollamaConfig, openrouterConfig, lmstudioConfig, aigatewayConfig, llamacppConfig, googleConfig];
 
 // Get the primary service (highest priority enabled service)
 export function getPrimaryService(): ServiceConfig | null {

src/config/services.ts

@@ -50,6 +50,14 @@ const serviceStatusSchema = z.object({
           model: z.string(),
           chatModel: z.string(),
         })
+      }),
+      google: z.object({
+        enabled: z.boolean(),
+        available: z.boolean(),
+        config: z.object({
+          model: z.string(),
+          hasApiKey: z.boolean(),
+        })
       })
   }),
   primary: z.string().nullable(),
@@ -62,7 +70,7 @@ const modelsSchema = z.object({
   available: z.boolean()
 })
 
-const capabilityEnum = z.enum(['summarize', 'rewrite', 'compose', 'keywords', 'sentiment', 'emailReply', 'vision', 'askText', 'translate', 'meetingNotes', 'planning', 'outline'])
+const capabilityEnum = z.enum(['summarize', 'pdf-summarizer', 'rewrite', 'compose', 'keywords', 'sentiment', 'emailReply', 'vision', 'askText', 'translate', 'meetingNotes', 'planning', 'outline'])
 const byProviderSchema = z.record(z.array(z.string()))
 const providerViewSchema = z.record(z.array(z.object({
   name: z.string(),
@@ -73,6 +81,7 @@ const providerViewSchema = z.record(z.array(z.object({
     source: z.literal('config'),
     byCapability: z.object({
       summarize: byProviderSchema,
+      'pdf-summarizer': byProviderSchema,
       rewrite: byProviderSchema,
       compose: byProviderSchema,
       keywords: byProviderSchema,
@@ -105,6 +114,7 @@ async function handleGetModels(c: Context) {
     if (source === 'config') {
       const byCapability = {
         summarize: getModelsByCapability('summarize'),
+        'pdf-summarizer': getModelsByCapability('pdf-summarizer'),
         rewrite: getModelsByCapability('rewrite'),
         compose: getModelsByCapability('compose'),
         keywords: getModelsByCapability('keywords'),
@@ -128,7 +138,8 @@ async function handleGetModels(c: Context) {
         getAvailableModels('ollama'),
         getAvailableModels('openrouter'),
         getAvailableModels('anthropic'),
-        getAvailableModels('lmstudio')
+        getAvailableModels('lmstudio'),
+        getAvailableModels('google')
       ])
       const response = {
         openai: {
@@ -155,6 +166,11 @@ async function handleGetModels(c: Context) {
           service: 'lmstudio',
           models: results[4].status === 'fulfilled' ? results[4].value : [],
           available: await checkServiceAvailability('lmstudio')
+        },
+        google: {
+          service: 'google',
+          models: results[5].status === 'fulfilled' ? results[5].value : [],
+          available: await checkServiceAvailability('google')
         }
       }
       return c.json(response, 200)
@@ -175,8 +191,8 @@ async function handleGetModels(c: Context) {
 async function handleServiceHealth(c: Context) {
   try {
     const service = c.req.param('service')
-    if (!service || !['openai', 'ollama', 'openrouter', 'anthropic', 'lmstudio'].includes(service)) {
-      return c.json({ error: 'Invalid service. Must be one of: openai, ollama, openrouter, anthropic, lmstudio' }, 400)
+    if (!service || !['openai', 'ollama', 'openrouter', 'anthropic', 'lmstudio', 'google'].includes(service)) {
+      return c.json({ error: 'Invalid service. Must be one of: openai, ollama, openrouter, anthropic, lmstudio, google' }, 400)
     }
     const available = await checkServiceAvailability(service as any)
     return c.json({
@@ -236,6 +252,7 @@ router.openapi(
                 openrouter: modelsSchema,
                 anthropic: modelsSchema,
                 lmstudio: modelsSchema,
+                google: modelsSchema,
               }),
               modelsGuidanceSchema.extend({ byProvider: providerViewSchema }).partial({ byCapability: true, byProvider: true })
             ])

src/routes/v1/services.ts

@@ -7,6 +7,7 @@ import {
   lmstudioConfig,
   aigatewayConfig,
   llamacppConfig,
+  googleConfig,
   isServiceEnabled 
 } from "../config/services";
 import { llmRequestSchema } from "../schemas/v1/llm";
@@ -21,6 +22,7 @@ enum Provider {
   lmstudio = 'lmstudio',
   aigateway = 'aigateway',
   llamacpp = 'llamacpp',
+  google = 'google',
 }
 
 // Service types
@@ -68,6 +70,8 @@ export async function checkServiceAvailability(service: AIService): Promise<bool
       return isServiceEnabled('AIGateway');
     case Provider.llamacpp:
       return isServiceEnabled('LlamaCpp');
+    case Provider.google:
+      return isServiceEnabled('Google');
     default:
       return false;
   }
@@ -179,6 +183,14 @@ export async function getServiceStatus() {
       config: {
         baseURL: llamacppConfig.baseURL,
       }
+    },
+    google: {
+      enabled: isServiceEnabled('Google'),
+      available: await checkServiceAvailability(Provider.google),
+      config: {
+        model: googleConfig.model,
+        hasApiKey: !!googleConfig.apiKey,
+      }
     }
   };