feat: tool-calling support across providers (#3)

README.md

@@ -69,6 +69,7 @@ The problem is that stacking them by hand is painful: fourteen different SDKs, f
 
 - **OpenAI-compatible** — `POST /v1/chat/completions` and `GET /v1/models` work with the official OpenAI SDKs and any OpenAI-compatible client (LangChain, LlamaIndex, Continue, Hermes, etc.). Just change `base_url`.
 - **Streaming and non-streaming** — Server-Sent Events for `stream: true`, JSON response otherwise. Every provider adapter implements both.
+- **Tool calling** — OpenAI-style `tools` / `tool_choice` requests are passed through, and assistant `tool_calls` + `tool` role follow-up messages round-trip across providers.
 - **Automatic fallover** — If the chosen provider returns a 429, 5xx, or times out, the router skips it, puts the key on a short cooldown, and retries on the next model in your fallback chain (up to 20 attempts).
 - **Per-key rate tracking** — RPM, RPD, TPM, and TPD counters per `(platform, model, key)` so the router always picks a key that's under its caps.
 - **Sticky sessions** — Multi-turn conversations keep talking to the same model for 30 minutes to avoid the hallucination spike that comes from mid-conversation model switches.
@@ -86,7 +87,6 @@ The scope is deliberately narrow. If a feature isn't on this list and isn't belo
 - **Embeddings** (`/v1/embeddings`)
 - **Image generation** (`/v1/images/*`)
 - **Audio / speech** (`/v1/audio/*`)
-- **Function / tool calling** — the request schema doesn't pass `tools` through yet
 - **Vision / multimodal inputs** — message content is text-only
 - **Legacy completions** (`/v1/completions`) — only the chat endpoint is implemented
 - **Moderation** (`/v1/moderations`)

server/src/__tests__/providers/cloudflare.test.ts

@@ -16,13 +16,22 @@ describe('CloudflareProvider', () => {
   it('should parse account_id:token key format', async () => {
     let capturedUrl = '';
     let capturedHeaders: Record<string, string> = {};
+    let capturedBody: any = null;
 
     vi.spyOn(global, 'fetch').mockImplementation(async (url, init) => {
       capturedUrl = url as string;
       capturedHeaders = (init as any).headers;
+      capturedBody = JSON.parse((init as any).body);
       return {
         ok: true,
-        json: () => Promise.resolve({ result: { response: 'Hello from CF!' } }),
+        json: () => Promise.resolve({
+          id: 'chatcmpl-cf',
+          object: 'chat.completion',
+          created: 123,
+          model: '@cf/meta/llama-3.1-70b-instruct',
+          choices: [{ index: 0, message: { role: 'assistant', content: 'Hello from CF!' }, finish_reason: 'stop' }],
+          usage: { prompt_tokens: 1, completion_tokens: 2, total_tokens: 3 },
+        }),
       } as any;
     });
 
@@ -33,8 +42,9 @@ describe('CloudflareProvider', () => {
     );
 
     expect(capturedUrl).toContain('abc123');
-    expect(capturedUrl).toContain('@cf/meta/llama-3.1-70b-instruct');
+    expect(capturedUrl).toContain('/ai/v1/chat/completions');
     expect(capturedHeaders['Authorization']).toBe('Bearer my-token-here');
+    expect(capturedBody.model).toBe('@cf/meta/llama-3.1-70b-instruct');
     expect(result.choices[0].message.content).toBe('Hello from CF!');
   });
 

server/src/__tests__/providers/cohere.test.ts

@@ -13,23 +13,45 @@ describe('CohereProvider', () => {
     expect(provider.name).toBe('Cohere');
   });
 
-  it('should translate response to OpenAI format', async () => {
-    vi.spyOn(global, 'fetch').mockResolvedValueOnce({
+  it('should call compatibility API and return OpenAI response', async () => {
+    let capturedUrl = '';
+    let capturedBody: any = null;
+    vi.spyOn(global, 'fetch').mockImplementationOnce(async (url, init) => {
+      capturedUrl = String(url);
+      capturedBody = JSON.parse((init as any).body);
+      return {
         ok: true,
         json: () => Promise.resolve({
           id: 'cohere-123',
-        message: { content: [{ type: 'text', text: 'Hello from Cohere!' }] },
-        finish_reason: 'COMPLETE',
-        usage: { tokens: { input_tokens: 10, output_tokens: 5 } },
+          object: 'chat.completion',
+          created: 123,
+          model: 'command-a-03-2025',
+          choices: [{ index: 0, message: { role: 'assistant', content: 'Hello from Cohere!' }, finish_reason: 'stop' }],
+          usage: { prompt_tokens: 10, completion_tokens: 5, total_tokens: 15 },
         }),
-    } as any);
+      } as any;
+    });
 
     const result = await provider.chatCompletion(
       'test-key',
       [{ role: 'user', content: 'Hi' }],
       'command-r-plus-08-2024',
+      {
+        tools: [{
+          type: 'function',
+          function: {
+            name: 'get_weather',
+            parameters: {
+              type: 'object',
+              properties: { city: { type: 'string' } },
+            },
+          },
+        }],
+      },
     );
 
+    expect(capturedUrl).toContain('/compatibility/v1/chat/completions');
+    expect(capturedBody.tools).toHaveLength(1);
     expect(result.object).toBe('chat.completion');
     expect(result.choices[0].message.content).toBe('Hello from Cohere!');
     expect(result.usage.prompt_tokens).toBe(10);

server/src/__tests__/providers/google.test.ts

@@ -92,4 +92,83 @@ describe('GoogleProvider', () => {
     expect(capturedBody.contents).toHaveLength(1);
     expect(capturedBody.contents[0].role).toBe('user');
   });
+
+  it('should translate OpenAI tools/tool_choice to Gemini tools/toolConfig', async () => {
+    let capturedBody: any;
+    vi.spyOn(global, 'fetch').mockImplementation(async (_url, init) => {
+      capturedBody = JSON.parse((init as any).body);
+      return {
+        ok: true,
+        json: () => Promise.resolve({
+          candidates: [{ content: { parts: [{ text: 'ok' }] }, finishReason: 'STOP' }],
+          usageMetadata: { promptTokenCount: 1, candidatesTokenCount: 1, totalTokenCount: 2 },
+        }),
+      } as any;
+    });
+
+    await provider.chatCompletion(
+      'test-key',
+      [{ role: 'user', content: 'Weather in Karachi?' }],
+      'gemini-2.5-pro',
+      {
+        tools: [{
+          type: 'function',
+          function: {
+            name: 'get_weather',
+            description: 'Get weather for a city',
+            parameters: {
+              type: 'object',
+              properties: { city: { type: 'string' } },
+              required: ['city'],
+            },
+          },
+        }],
+        tool_choice: {
+          type: 'function',
+          function: { name: 'get_weather' },
+        },
+      },
+    );
+
+    expect(capturedBody.tools[0].functionDeclarations[0].name).toBe('get_weather');
+    expect(capturedBody.toolConfig.functionCallingConfig.mode).toBe('ANY');
+    expect(capturedBody.toolConfig.functionCallingConfig.allowedFunctionNames).toEqual(['get_weather']);
+  });
+
+  it('should translate Gemini functionCall response to OpenAI tool_calls', async () => {
+    vi.spyOn(global, 'fetch').mockResolvedValueOnce({
+      ok: true,
+      json: () => Promise.resolve({
+        candidates: [{
+          content: {
+            parts: [{
+              functionCall: {
+                id: 'call_123',
+                name: 'get_weather',
+                args: { city: 'Lahore' },
+              },
+            }],
+          },
+          finishReason: 'STOP',
+        }],
+        usageMetadata: {
+          promptTokenCount: 12,
+          candidatesTokenCount: 3,
+          totalTokenCount: 15,
+        },
+      }),
+    } as any);
+
+    const result = await provider.chatCompletion(
+      'test-key',
+      [{ role: 'user', content: 'What is the weather?' }],
+      'gemini-2.5-pro',
+    );
+
+    expect(result.choices[0].finish_reason).toBe('tool_calls');
+    expect(result.choices[0].message.content).toBeNull();
+    expect(result.choices[0].message.tool_calls?.[0].id).toBe('call_123');
+    expect(result.choices[0].message.tool_calls?.[0].function.name).toBe('get_weather');
+    expect(result.choices[0].message.tool_calls?.[0].function.arguments).toBe('{"city":"Lahore"}');
+  });
 });

server/src/__tests__/providers/openai-compat.test.ts

@@ -21,10 +21,12 @@ describe('OpenAICompatProvider', () => {
   it('should call API with correct URL and headers', async () => {
     let capturedUrl = '';
     let capturedHeaders: Record<string, string> = {};
+    let capturedBody: any = null;
 
     vi.spyOn(global, 'fetch').mockImplementation(async (url, init) => {
       capturedUrl = url as string;
       capturedHeaders = (init as any).headers;
+      capturedBody = JSON.parse((init as any).body);
       return {
         ok: true,
         json: () => Promise.resolve({
@@ -43,6 +45,51 @@ describe('OpenAICompatProvider', () => {
     expect(capturedUrl).toBe('https://api.test.com/v1/chat/completions');
     expect(capturedHeaders['Authorization']).toBe('Bearer my-key');
     expect(capturedHeaders['X-Custom']).toBe('test');
+    expect(capturedBody.messages[0].role).toBe('user');
+  });
+
+  it('should pass tool-calling params through untouched', async () => {
+    let capturedBody: any = null;
+    vi.spyOn(global, 'fetch').mockImplementation(async (_url, init) => {
+      capturedBody = JSON.parse((init as any).body);
+      return {
+        ok: true,
+        json: () => Promise.resolve({
+          id: 'test-id',
+          object: 'chat.completion',
+          created: 123,
+          model: 'test-model',
+          choices: [{ index: 0, message: { role: 'assistant', content: null, tool_calls: [] }, finish_reason: 'stop' }],
+          usage: { prompt_tokens: 1, completion_tokens: 1, total_tokens: 2 },
+        }),
+      } as any;
+    });
+
+    await provider.chatCompletion(
+      'my-key',
+      [{ role: 'user', content: 'what is weather?' }],
+      'test-model',
+      {
+        tools: [{
+          type: 'function',
+          function: {
+            name: 'get_weather',
+            description: 'Get weather',
+            parameters: {
+              type: 'object',
+              properties: { city: { type: 'string' } },
+              required: ['city'],
+            },
+          },
+        }],
+        tool_choice: 'required',
+        parallel_tool_calls: true,
+      },
+    );
+
+    expect(capturedBody.tools).toHaveLength(1);
+    expect(capturedBody.tool_choice).toBe('required');
+    expect(capturedBody.parallel_tool_calls).toBe(true);
   });
 
   it('should throw on error response', async () => {

server/src/__tests__/routes/proxy-tools.test.ts

+import { describe, it, expect, beforeAll, beforeEach, afterEach, vi } from 'vitest';
+import type { Express } from 'express';
+import { createApp } from '../../app.js';
+import { initDb, getDb } from '../../db/index.js';
+
+async function request(app: Express, method: string, path: string, body?: any) {
+  const server = app.listen(0);
+  const addr = server.address() as any;
+  const url = `http://127.0.0.1:${addr.port}${path}`;
+
+  const res = await fetch(url, {
+    method,
+    headers: body ? { 'Content-Type': 'application/json' } : {},
+    body: body ? JSON.stringify(body) : undefined,
+  });
+
+  const data = await res.text();
+  server.close();
+
+  let json: any = null;
+  try { json = JSON.parse(data); } catch {}
+
+  return { status: res.status, body: json, headers: res.headers, raw: data };
+}
+
+describe('Proxy tool-calling support', () => {
+  let app: Express;
+
+  beforeAll(() => {
+    process.env.ENCRYPTION_KEY = '0'.repeat(64);
+    initDb(':memory:');
+    app = createApp();
+  });
+
+  beforeEach(async () => {
+    const db = getDb();
+    db.prepare('DELETE FROM api_keys').run();
+    db.prepare('DELETE FROM requests').run();
+
+    const addKey = await request(app, 'POST', '/api/keys', {
+      platform: 'groq',
+      key: 'gsk_proxy_tool_test',
+      label: 'proxy-tools',
+    });
+    expect(addKey.status).toBe(201);
+  });
+
+  afterEach(() => {
+    vi.restoreAllMocks();
+  });
+
+  it('passes tools/tool_choice to provider and returns tool_calls', async () => {
+    const origFetch = global.fetch;
+    let providerBody: any = null;
+
+    vi.spyOn(global, 'fetch').mockImplementation(async (url, init) => {
+      const urlStr = typeof url === 'string' ? url : url.toString();
+      if (urlStr.includes('api.groq.com/openai/v1/chat/completions')) {
+        providerBody = JSON.parse((init as any).body);
+        return {
+          ok: true,
+          json: () => Promise.resolve({
+            id: 'chatcmpl-tool',
+            object: 'chat.completion',
+            created: 123,
+            model: 'openai/gpt-oss-120b',
+            choices: [{
+              index: 0,
+              message: {
+                role: 'assistant',
+                content: null,
+                tool_calls: [{
+                  id: 'call_weather',
+                  type: 'function',
+                  function: {
+                    name: 'get_weather',
+                    arguments: '{"city":"Karachi"}',
+                  },
+                }],
+              },
+              finish_reason: 'tool_calls',
+            }],
+            usage: { prompt_tokens: 12, completion_tokens: 4, total_tokens: 16 },
+          }),
+        } as any;
+      }
+      return origFetch(url, init);
+    });
+
+    const { status, body } = await request(app, 'POST', '/v1/chat/completions', {
+      model: 'auto',
+      messages: [{ role: 'user', content: 'What is the weather in Karachi?' }],
+      tools: [{
+        type: 'function',
+        function: {
+          name: 'get_weather',
+          description: 'Get current weather',
+          parameters: {
+            type: 'object',
+            properties: { city: { type: 'string' } },
+            required: ['city'],
+          },
+        },
+      }],
+      tool_choice: 'required',
+    });
+
+    expect(status).toBe(200);
+    expect(providerBody.tools).toHaveLength(1);
+    expect(providerBody.tool_choice).toBe('required');
+    expect(body.choices[0].finish_reason).toBe('tool_calls');
+    expect(body.choices[0].message.tool_calls[0].function.name).toBe('get_weather');
+  });
+
+  it('accepts assistant tool_calls + tool messages in follow-up turns', async () => {
+    const origFetch = global.fetch;
+    let providerBody: any = null;
+
+    vi.spyOn(global, 'fetch').mockImplementation(async (url, init) => {
+      const urlStr = typeof url === 'string' ? url : url.toString();
+      if (urlStr.includes('api.groq.com/openai/v1/chat/completions')) {
+        providerBody = JSON.parse((init as any).body);
+        return {
+          ok: true,
+          json: () => Promise.resolve({
+            id: 'chatcmpl-final',
+            object: 'chat.completion',
+            created: 123,
+            model: 'openai/gpt-oss-120b',
+            choices: [{
+              index: 0,
+              message: {
+                role: 'assistant',
+                content: 'It is 30C in Karachi.',
+              },
+              finish_reason: 'stop',
+            }],
+            usage: { prompt_tokens: 18, completion_tokens: 6, total_tokens: 24 },
+          }),
+        } as any;
+      }
+      return origFetch(url, init);
+    });
+
+    const { status, body } = await request(app, 'POST', '/v1/chat/completions', {
+      messages: [
+        { role: 'user', content: 'Weather in Karachi?' },
+        {
+          role: 'assistant',
+          content: null,
+          tool_calls: [{
+            id: 'call_weather_1',
+            type: 'function',
+            function: {
+              name: 'get_weather',
+              arguments: '{"city":"Karachi"}',
+            },
+          }],
+        },
+        {
+          role: 'tool',
+          tool_call_id: 'call_weather_1',
+          content: '{"temp_c":30}',
+        },
+      ],
+    });
+
+    expect(status).toBe(200);
+    expect(providerBody.messages[1].role).toBe('assistant');
+    expect(providerBody.messages[1].content).toBeNull();
+    expect(providerBody.messages[1].tool_calls).toHaveLength(1);
+    expect(providerBody.messages[2].role).toBe('tool');
+    expect(providerBody.messages[2].tool_call_id).toBe('call_weather_1');
+    expect(body.choices[0].message.content).toContain('30C');
+  });
+});

server/src/providers/base.ts

@@ -2,6 +2,8 @@ import type {
   ChatMessage,
   ChatCompletionResponse,
   ChatCompletionChunk,
+  ChatToolDefinition,
+  ChatToolChoice,
   Platform,
 } from '@freellmapi/shared/types.js';
 
@@ -10,6 +12,9 @@ export interface CompletionOptions {
   temperature?: number;
   max_tokens?: number;
   top_p?: number;
+  tools?: ChatToolDefinition[];
+  tool_choice?: ChatToolChoice;
+  parallel_tool_calls?: boolean;
 }
 
 export abstract class BaseProvider {

server/src/providers/cloudflare.ts

@@ -27,7 +27,7 @@ export class CloudflareProvider extends BaseProvider {
     options?: CompletionOptions,
   ): Promise<ChatCompletionResponse> {
     const { accountId, token } = this.parseKey(apiKey);
-    const url = `https://api.cloudflare.com/client/v4/accounts/${accountId}/ai/run/${modelId}`;
+    const url = `https://api.cloudflare.com/client/v4/accounts/${accountId}/ai/v1/chat/completions`;
 
     const res = await this.fetchWithTimeout(url, {
       method: 'POST',
@@ -36,38 +36,25 @@ export class CloudflareProvider extends BaseProvider {
         'Content-Type': 'application/json',
       },
       body: JSON.stringify({
+        model: modelId,
         messages,
-        max_tokens: options?.max_tokens,
         temperature: options?.temperature,
+        max_tokens: options?.max_tokens,
+        top_p: options?.top_p,
+        tools: options?.tools,
+        tool_choice: options?.tool_choice,
+        parallel_tool_calls: options?.parallel_tool_calls,
       }),
     });
 
     if (!res.ok) {
       const err = await res.json().catch(() => ({}));
-      const errors = (err as any).errors;
-      throw new Error(`Cloudflare API error ${res.status}: ${errors?.[0]?.message ?? res.statusText}`);
+      throw new Error(`Cloudflare API error ${res.status}: ${(err as any).error?.message ?? (err as any).errors?.[0]?.message ?? res.statusText}`);
     }
 
-    const data = await res.json() as any;
-    const text = data.result?.response ?? '';
-
-    return {
-      id: this.makeId(),
-      object: 'chat.completion',
-      created: Math.floor(Date.now() / 1000),
-      model: modelId,
-      choices: [{
-        index: 0,
-        message: { role: 'assistant', content: text },
-        finish_reason: 'stop',
-      }],
-      usage: {
-        prompt_tokens: 0,
-        completion_tokens: 0,
-        total_tokens: 0,
-      },
-      _routed_via: { platform: 'cloudflare', model: modelId },
-    };
+    const data = await res.json() as ChatCompletionResponse;
+    data._routed_via = { platform: 'cloudflare', model: modelId };
+    return data;
   }
 
   async *streamChatCompletion(
@@ -77,7 +64,7 @@ export class CloudflareProvider extends BaseProvider {
     options?: CompletionOptions,
   ): AsyncGenerator<ChatCompletionChunk> {
     const { accountId, token } = this.parseKey(apiKey);
-    const url = `https://api.cloudflare.com/client/v4/accounts/${accountId}/ai/run/${modelId}`;
+    const url = `https://api.cloudflare.com/client/v4/accounts/${accountId}/ai/v1/chat/completions`;
 
     const res = await this.fetchWithTimeout(url, {
       method: 'POST',
@@ -86,23 +73,27 @@ export class CloudflareProvider extends BaseProvider {
         'Content-Type': 'application/json',
       },
       body: JSON.stringify({
+        model: modelId,
         messages,
-        max_tokens: options?.max_tokens,
         temperature: options?.temperature,
+        max_tokens: options?.max_tokens,
+        top_p: options?.top_p,
+        tools: options?.tools,
+        tool_choice: options?.tool_choice,
+        parallel_tool_calls: options?.parallel_tool_calls,
         stream: true,
       }),
     });
 
     if (!res.ok) {
       const err = await res.json().catch(() => ({}));
-      throw new Error(`Cloudflare API error ${res.status}: ${(err as any).errors?.[0]?.message ?? res.statusText}`);
+      throw new Error(`Cloudflare API error ${res.status}: ${(err as any).error?.message ?? (err as any).errors?.[0]?.message ?? res.statusText}`);
     }
 
     const reader = res.body?.getReader();
     if (!reader) throw new Error('No response body');
 
     const decoder = new TextDecoder();
-    const id = this.makeId();
     let buffer = '';
 
     while (true) {
@@ -119,17 +110,10 @@ export class CloudflareProvider extends BaseProvider {
         const data = trimmed.slice(6);
         if (data === '[DONE]') return;
         try {
-          const parsed = JSON.parse(data);
-          if (parsed.response) {
-            yield {
-              id,
-              object: 'chat.completion.chunk',
-              created: Math.floor(Date.now() / 1000),
-              model: modelId,
-              choices: [{ index: 0, delta: { content: parsed.response }, finish_reason: null }],
-            };
+          yield JSON.parse(data) as ChatCompletionChunk;
+        } catch {
+          // Skip malformed chunks
         }
-        } catch { /* skip */ }
       }
     }
   }

server/src/providers/cohere.ts

@@ -5,16 +5,7 @@ import type {
 } from '@freellmapi/shared/types.js';
 import { BaseProvider, type CompletionOptions } from './base.js';
 
-const API_BASE = 'https://api.cohere.com/v2';
-
-interface CohereResponse {
-  id: string;
-  message?: { content?: { type: string; text: string }[] };
-  finish_reason?: string;
-  usage?: {
-    tokens?: { input_tokens?: number; output_tokens?: number };
-  };
-}
+const API_BASE = 'https://api.cohere.ai/compatibility/v1';
 
 export class CohereProvider extends BaseProvider {
   readonly platform = 'cohere' as const;
@@ -26,51 +17,33 @@ export class CohereProvider extends BaseProvider {
     modelId: string,
     options?: CompletionOptions,
   ): Promise<ChatCompletionResponse> {
-    const cohereMessages = messages.map(m => ({
-      role: m.role === 'system' ? 'system' as const : m.role === 'assistant' ? 'assistant' as const : 'user' as const,
-      content: m.content,
-    }));
+    const body: Record<string, unknown> = {
+      model: modelId,
+      messages,
+      temperature: options?.temperature,
+      max_tokens: options?.max_tokens,
+      top_p: options?.top_p,
+      tools: options?.tools,
+      tool_choice: options?.tool_choice,
+    };
 
-    const res = await this.fetchWithTimeout(`${API_BASE}/chat`, {
+    const res = await this.fetchWithTimeout(`${API_BASE}/chat/completions`, {
       method: 'POST',
       headers: {
         'Authorization': `Bearer ${apiKey}`,
         'Content-Type': 'application/json',
       },
-      body: JSON.stringify({
-        model: modelId,
-        messages: cohereMessages,
-        temperature: options?.temperature,
-        max_tokens: options?.max_tokens,
-        p: options?.top_p,
-      }),
+      body: JSON.stringify(body),
     });
 
     if (!res.ok) {
       const err = await res.json().catch(() => ({}));
-      throw new Error(`Cohere API error ${res.status}: ${(err as any).message ?? res.statusText}`);
+      throw new Error(`Cohere API error ${res.status}: ${(err as any).error?.message ?? res.statusText}`);
     }
 
-    const data = await res.json() as CohereResponse;
-    const text = data.message?.content?.[0]?.text ?? '';
-
-    return {
-      id: data.id ?? this.makeId(),
-      object: 'chat.completion',
-      created: Math.floor(Date.now() / 1000),
-      model: modelId,
-      choices: [{
-        index: 0,
-        message: { role: 'assistant', content: text },
-        finish_reason: data.finish_reason ?? 'stop',
-      }],
-      usage: {
-        prompt_tokens: data.usage?.tokens?.input_tokens ?? 0,
-        completion_tokens: data.usage?.tokens?.output_tokens ?? 0,
-        total_tokens: (data.usage?.tokens?.input_tokens ?? 0) + (data.usage?.tokens?.output_tokens ?? 0),
-      },
-      _routed_via: { platform: 'cohere', model: modelId },
-    };
+    const data = await res.json() as ChatCompletionResponse;
+    data._routed_via = { platform: 'cohere', model: modelId };
+    return data;
   }
 
   async *streamChatCompletion(
@@ -79,36 +52,35 @@ export class CohereProvider extends BaseProvider {
     modelId: string,
     options?: CompletionOptions,
   ): AsyncGenerator<ChatCompletionChunk> {
-    const cohereMessages = messages.map(m => ({
-      role: m.role === 'system' ? 'system' as const : m.role === 'assistant' ? 'assistant' as const : 'user' as const,
-      content: m.content,
-    }));
+    const body: Record<string, unknown> = {
+      model: modelId,
+      messages,
+      temperature: options?.temperature,
+      max_tokens: options?.max_tokens,
+      top_p: options?.top_p,
+      tools: options?.tools,
+      tool_choice: options?.tool_choice,
+      stream: true,
+    };
 
-    const res = await this.fetchWithTimeout(`${API_BASE}/chat`, {
+    const res = await this.fetchWithTimeout(`${API_BASE}/chat/completions`, {
       method: 'POST',
       headers: {
         'Authorization': `Bearer ${apiKey}`,
         'Content-Type': 'application/json',
       },
-      body: JSON.stringify({
-        model: modelId,
-        messages: cohereMessages,
-        temperature: options?.temperature,
-        max_tokens: options?.max_tokens,
-        stream: true,
-      }),
+      body: JSON.stringify(body),
     });
 
     if (!res.ok) {
       const err = await res.json().catch(() => ({}));
-      throw new Error(`Cohere API error ${res.status}: ${(err as any).message ?? res.statusText}`);
+      throw new Error(`Cohere API error ${res.status}: ${(err as any).error?.message ?? res.statusText}`);
     }
 
     const reader = res.body?.getReader();
     if (!reader) throw new Error('No response body');
 
     const decoder = new TextDecoder();
-    const id = this.makeId();
     let buffer = '';
 
     while (true) {
@@ -121,31 +93,13 @@ export class CohereProvider extends BaseProvider {
 
       for (const line of lines) {
         const trimmed = line.trim();
-        if (!trimmed) continue;
+        if (!trimmed || !trimmed.startsWith('data: ')) continue;
+        const data = trimmed.slice(6);
+        if (data === '[DONE]') return;
         try {
-          const event = JSON.parse(trimmed);
-          if (event.type === 'content-delta') {
-            const text = event.delta?.message?.content?.text ?? '';
-            if (text) {
-              yield {
-                id,
-                object: 'chat.completion.chunk',
-                created: Math.floor(Date.now() / 1000),
-                model: modelId,
-                choices: [{ index: 0, delta: { content: text }, finish_reason: null }],
-              };
-            }
-          } else if (event.type === 'message-end') {
-            yield {
-              id,
-              object: 'chat.completion.chunk',
-              created: Math.floor(Date.now() / 1000),
-              model: modelId,
-              choices: [{ index: 0, delta: {}, finish_reason: 'stop' }],
-            };
-          }
+          yield JSON.parse(data) as ChatCompletionChunk;
         } catch {
-          // Skip malformed lines
+          // Skip malformed chunks
         }
       }
     }

server/src/providers/google.ts

@@ -2,40 +2,205 @@ import type {
   ChatMessage,
   ChatCompletionResponse,
   ChatCompletionChunk,
+  ChatToolCall,
+  ChatToolChoice,
+  ChatToolDefinition,
   TokenUsage,
 } from '@freellmapi/shared/types.js';
 import { BaseProvider, type CompletionOptions } from './base.js';
 
 const API_BASE = 'https://generativelanguage.googleapis.com/v1beta';
 
+interface GeminiPart {
+  text?: string;
+  functionCall?: {
+    id?: string;
+    name?: string;
+    args?: unknown;
+  };
+  functionResponse?: {
+    id?: string;
+    name?: string;
+    response?: unknown;
+  };
+}
+
+interface GeminiCandidate {
+  content?: { parts?: GeminiPart[] };
+  finishReason?: string;
+}
+
+interface GeminiResponse {
+  candidates?: GeminiCandidate[];
+  usageMetadata?: {
+    promptTokenCount?: number;
+    candidatesTokenCount?: number;
+    totalTokenCount?: number;
+  };
+}
+
+function safeParseObject(raw: string): Record<string, unknown> {
+  try {
+    const parsed = JSON.parse(raw) as unknown;
+    if (parsed && typeof parsed === 'object' && !Array.isArray(parsed)) {
+      return parsed as Record<string, unknown>;
+    }
+    return { value: parsed };
+  } catch {
+    return { value: raw };
+  }
+}
+
+function normalizeGeminiArgs(args: unknown): string {
+  if (typeof args === 'string') return args;
+  return JSON.stringify(args ?? {});
+}
+
+function toGeminiFinishReason(finishReason?: string): string {
+  const r = (finishReason ?? '').toUpperCase();
+  if (!r) return 'stop';
+  if (r === 'MAX_TOKENS') return 'length';
+  if (r === 'SAFETY' || r === 'RECITATION' || r === 'BLOCKLIST' || r === 'PROHIBITED_CONTENT' || r === 'SPII') {
+    return 'content_filter';
+  }
+  return 'stop';
+}
+
+function toGeminiTools(tools?: ChatToolDefinition[]): Array<{ functionDeclarations: Array<Record<string, unknown>> }> | undefined {
+  if (!tools || tools.length === 0) return undefined;
+
+  return [{
+    functionDeclarations: tools.map(t => ({
+      name: t.function.name,
+      description: t.function.description,
+      parameters: t.function.parameters,
+    })),
+  }];
+}
+
+function toGeminiToolConfig(toolChoice?: ChatToolChoice): { functionCallingConfig: Record<string, unknown> } | undefined {
+  if (!toolChoice) return undefined;
+
+  if (typeof toolChoice === 'string') {
+    const mode =
+      toolChoice === 'none'
+        ? 'NONE'
+        : toolChoice === 'required'
+          ? 'ANY'
+          : 'AUTO';
+    return { functionCallingConfig: { mode } };
+  }
+
+  return {
+    functionCallingConfig: {
+      mode: 'ANY',
+      allowedFunctionNames: [toolChoice.function.name],
+    },
+  };
+}
+
 // Translate OpenAI messages to Gemini format
 function toGeminiContents(messages: ChatMessage[]) {
-  const systemInstruction = messages.find(m => m.role === 'system');
+  const systemMessages = messages
+    .filter(m => m.role === 'system' && typeof m.content === 'string' && m.content.length > 0)
+    .map(m => m.content as string);
+
+  const toolNameByCallId = new Map<string, string>();
+  for (const m of messages) {
+    for (const tc of m.tool_calls ?? []) {
+      toolNameByCallId.set(tc.id, tc.function.name);
+    }
+  }
+
   const contents = messages
     .filter(m => m.role !== 'system')
-    .map(m => ({
-      role: m.role === 'assistant' ? 'model' : 'user',
-      parts: [{ text: m.content }],
-    }));
+    .map((m): { role: 'user' | 'model'; parts: GeminiPart[] } | null => {
+      if (m.role === 'assistant') {
+        const parts: GeminiPart[] = [];
+
+        if (typeof m.content === 'string' && m.content.length > 0) {
+          parts.push({ text: m.content });
+        }
+
+        for (const call of m.tool_calls ?? []) {
+          parts.push({
+            functionCall: {
+              id: call.id,
+              name: call.function.name,
+              args: safeParseObject(call.function.arguments),
+            },
+          });
+        }
+
+        if (parts.length === 0) return null;
+        return {
+          role: 'model',
+          parts,
+        };
+      }
+
+      if (m.role === 'tool') {
+        const toolCallId = m.tool_call_id;
+        if (!toolCallId) return null;
+
+        const toolName = m.name ?? toolNameByCallId.get(toolCallId) ?? 'tool';
+        const response = safeParseObject(typeof m.content === 'string' ? m.content : '');
+
+        return {
+          role: 'user',
+          parts: [{
+            functionResponse: {
+              id: toolCallId,
+              name: toolName,
+              response,
+            },
+          }],
+        };
+      }
+
+      return {
+        role: 'user',
+        parts: [{ text: typeof m.content === 'string' ? m.content : '' }],
+      };
+    })
+    .filter((entry): entry is { role: 'user' | 'model'; parts: GeminiPart[] } => entry !== null);
 
   return {
     contents,
-    systemInstruction: systemInstruction
-      ? { parts: [{ text: systemInstruction.content }] }
+    systemInstruction: systemMessages.length > 0
+      ? { parts: [{ text: systemMessages.join('\n\n') }] }
       : undefined,
   };
 }
 
-interface GeminiResponse {
-  candidates?: {
-    content?: { parts?: { text?: string }[] };
-    finishReason?: string;
-  }[];
-  usageMetadata?: {
-    promptTokenCount?: number;
-    candidatesTokenCount?: number;
-    totalTokenCount?: number;
-  };
+function extractToolCalls(parts: GeminiPart[] | undefined): ChatToolCall[] {
+  const calls: ChatToolCall[] = [];
+  if (!parts) return calls;
+
+  let fallbackIndex = 0;
+  for (const part of parts) {
+    if (!part.functionCall?.name) continue;
+
+    const id = part.functionCall.id ?? `call_${Date.now()}_${fallbackIndex++}`;
+    calls.push({
+      id,
+      type: 'function',
+      function: {
+        name: part.functionCall.name,
+        arguments: normalizeGeminiArgs(part.functionCall.args),
+      },
+    });
+  }
+
+  return calls;
+}
+
+function extractText(parts: GeminiPart[] | undefined): string | null {
+  if (!parts) return null;
+  const text = parts
+    .map(p => p.text ?? '')
+    .join('');
+  return text.length > 0 ? text : null;
 }
 
 export class GoogleProvider extends BaseProvider {
@@ -57,6 +222,8 @@ export class GoogleProvider extends BaseProvider {
         maxOutputTokens: options?.max_tokens,
         topP: options?.top_p,
       },
+      tools: toGeminiTools(options?.tools),
+      toolConfig: toGeminiToolConfig(options?.tool_choice),
     };
     if (systemInstruction) body.systemInstruction = systemInstruction;
 
@@ -73,8 +240,11 @@ export class GoogleProvider extends BaseProvider {
     }
 
     const data = await res.json() as GeminiResponse;
+    const candidate = data.candidates?.[0];
+    const parts = candidate?.content?.parts;
+    const toolCalls = extractToolCalls(parts);
+    const text = extractText(parts);
 
-    const text = data.candidates?.[0]?.content?.parts?.[0]?.text ?? '';
     const usage: TokenUsage = {
       prompt_tokens: data.usageMetadata?.promptTokenCount ?? 0,
       completion_tokens: data.usageMetadata?.candidatesTokenCount ?? 0,
@@ -88,8 +258,12 @@ export class GoogleProvider extends BaseProvider {
       model: modelId,
       choices: [{
         index: 0,
-        message: { role: 'assistant', content: text },
-        finish_reason: data.candidates?.[0]?.finishReason?.toLowerCase() === 'stop' ? 'stop' : 'stop',
+        message: {
+          role: 'assistant',
+          content: text,
+          ...(toolCalls.length > 0 ? { tool_calls: toolCalls } : {}),
+        },
+        finish_reason: toolCalls.length > 0 ? 'tool_calls' : toGeminiFinishReason(candidate?.finishReason),
       }],
       usage,
       _routed_via: { platform: 'google', model: modelId },
@@ -111,6 +285,8 @@ export class GoogleProvider extends BaseProvider {
         maxOutputTokens: options?.max_tokens,
         topP: options?.top_p,
       },
+      tools: toGeminiTools(options?.tools),
+      toolConfig: toGeminiToolConfig(options?.tool_choice),
     };
     if (systemInstruction) body.systemInstruction = systemInstruction;
 
@@ -132,6 +308,10 @@ export class GoogleProvider extends BaseProvider {
     const decoder = new TextDecoder();
     const id = this.makeId();
     let buffer = '';
+    let emittedFinish = false;
+    let sawToolCalls = false;
+
+    const seenToolCallKeys = new Set<string>();
 
     while (true) {
       const { done, value } = await reader.read();
@@ -145,12 +325,38 @@ export class GoogleProvider extends BaseProvider {
         const trimmed = line.trim();
         if (!trimmed || !trimmed.startsWith('data: ')) continue;
         const raw = trimmed.slice(6);
-        if (raw === '[DONE]') return;
+        if (raw === '[DONE]') {
+          if (!emittedFinish) {
+            emittedFinish = true;
+            yield {
+              id,
+              object: 'chat.completion.chunk',
+              created: Math.floor(Date.now() / 1000),
+              model: modelId,
+              choices: [{
+                index: 0,
+                delta: {},
+                finish_reason: sawToolCalls ? 'tool_calls' : 'stop',
+              }],
+            };
+          }
+          return;
+        }
 
         const chunk = JSON.parse(raw) as GeminiResponse;
-        const text = chunk.candidates?.[0]?.content?.parts?.[0]?.text ?? '';
-        if (!text) continue;
+        const candidate = chunk.candidates?.[0];
+        const parts = candidate?.content?.parts ?? [];
 
+        const text = extractText(parts);
+        const toolCalls = extractToolCalls(parts).filter(call => {
+          const key = `${call.id}:${call.function.name}:${call.function.arguments}`;
+          if (seenToolCallKeys.has(key)) return false;
+          seenToolCallKeys.add(key);
+          return true;
+        });
+
+        if ((text && text.length > 0) || toolCalls.length > 0) {
+          sawToolCalls = sawToolCalls || toolCalls.length > 0;
           yield {
             id,
             object: 'chat.completion.chunk',
@@ -158,14 +364,34 @@ export class GoogleProvider extends BaseProvider {
             model: modelId,
             choices: [{
               index: 0,
-            delta: { content: text },
+              delta: {
+                ...(text ? { content: text } : {}),
+                ...(toolCalls.length > 0 ? { tool_calls: toolCalls } : {}),
+              },
               finish_reason: null,
             }],
           };
         }
+
+        if (candidate?.finishReason && !emittedFinish) {
+          emittedFinish = true;
+          yield {
+            id,
+            object: 'chat.completion.chunk',
+            created: Math.floor(Date.now() / 1000),
+            model: modelId,
+            choices: [{
+              index: 0,
+              delta: {},
+              finish_reason: sawToolCalls ? 'tool_calls' : toGeminiFinishReason(candidate.finishReason),
+            }],
+          };
+          return;
+        }
+      }
     }
 
-    // Final chunk
+    if (!emittedFinish) {
       yield {
         id,
         object: 'chat.completion.chunk',
@@ -174,10 +400,11 @@ export class GoogleProvider extends BaseProvider {
         choices: [{
           index: 0,
           delta: {},
-        finish_reason: 'stop',
+          finish_reason: sawToolCalls ? 'tool_calls' : 'stop',
         }],
       };
     }
+  }
 
   async validateKey(apiKey: string): Promise<boolean> {
     try {

server/src/providers/huggingface.ts

@@ -30,6 +30,9 @@ export class HuggingFaceProvider extends BaseProvider {
         temperature: options?.temperature,
         max_tokens: options?.max_tokens,
         top_p: options?.top_p,
+        tools: options?.tools,
+        tool_choice: options?.tool_choice,
+        parallel_tool_calls: options?.parallel_tool_calls,
       }),
     });
 
@@ -60,6 +63,10 @@ export class HuggingFaceProvider extends BaseProvider {
         messages,
         temperature: options?.temperature,
         max_tokens: options?.max_tokens,
+        top_p: options?.top_p,
+        tools: options?.tools,
+        tool_choice: options?.tool_choice,
+        parallel_tool_calls: options?.parallel_tool_calls,
         stream: true,
       }),
     });

server/src/providers/index.ts

@@ -68,10 +68,10 @@ register(new OpenAICompatProvider({
   baseUrl: 'https://models.inference.ai.azure.com',
 }));
 
-// Cohere - unique API format
+// Cohere - OpenAI-compatible via Cohere compatibility endpoint
 register(new CohereProvider());
 
-// Cloudflare Workers AI - unique API format (key = "account_id:token")
+// Cloudflare Workers AI - OpenAI-compatible endpoint (key = "account_id:token")
 register(new CloudflareProvider());
 
 // Hugging Face - OpenAI-compatible per-model endpoint

server/src/providers/openai-compat.ts

@@ -52,6 +52,9 @@ export class OpenAICompatProvider extends BaseProvider {
         temperature: options?.temperature,
         max_tokens: options?.max_tokens,
         top_p: options?.top_p,
+        tools: options?.tools,
+        tool_choice: options?.tool_choice,
+        parallel_tool_calls: options?.parallel_tool_calls,
       }),
     });
 
@@ -84,6 +87,9 @@ export class OpenAICompatProvider extends BaseProvider {
         temperature: options?.temperature,
         max_tokens: options?.max_tokens,
         top_p: options?.top_p,
+        tools: options?.tools,
+        tool_choice: options?.tool_choice,
+        parallel_tool_calls: options?.parallel_tool_calls,
         stream: true,
       }),
     });

server/src/routes/proxy.ts

 import { Router } from 'express';
 import type { Request, Response } from 'express';
 import { z } from 'zod';
+import type { ChatMessage } from '@freellmapi/shared/types.js';
 import { routeRequest, recordRateLimitHit, recordSuccess, type RouteResult } from '../services/router.js';
 import { recordRequest, recordTokens, setCooldown } from '../services/ratelimit.js';
 import { getDb, getUnifiedApiKey } from '../db/index.js';
@@ -13,16 +14,16 @@ export const proxyRouter = Router();
 const stickySessionMap = new Map<string, { modelDbId: number; lastUsed: number }>();
 const STICKY_TTL_MS = 30 * 60 * 1000; // 30 min session TTL
 
-function getSessionKey(messages: { role: string; content: string }[]): string {
+function getSessionKey(messages: ChatMessage[]): string {
   // Use the first user message as session identifier
   // Hermes sends the full conversation each time, so first user msg is stable
   const firstUser = messages.find(m => m.role === 'user');
-  if (!firstUser) return '';
+  if (!firstUser || typeof firstUser.content !== 'string') return '';
   // Hash: first 100 chars of first user message + message count
   return `${firstUser.content.slice(0, 100)}:${messages.length > 2 ? 'multi' : 'single'}`;
 }
 
-function getStickyModel(messages: { role: string; content: string }[]): number | undefined {
+function getStickyModel(messages: ChatMessage[]): number | undefined {
   // Only apply sticky for multi-turn (has assistant messages = continuation)
   const hasAssistant = messages.some(m => m.role === 'assistant');
   if (!hasAssistant) return undefined;
@@ -40,7 +41,7 @@ function getStickyModel(messages: { role: string; content: string }[]): number |
   return entry.modelDbId;
 }
 
-function setStickyModel(messages: { role: string; content: string }[], modelDbId: number) {
+function setStickyModel(messages: ChatMessage[], modelDbId: number) {
   const key = getSessionKey(messages);
   if (!key) return;
   stickySessionMap.set(key, { modelDbId, lastUsed: Date.now() });
@@ -73,16 +74,82 @@ proxyRouter.get('/models', (_req: Request, res: Response) => {
 
 const MAX_RETRIES = 20;
 
-const chatCompletionSchema = z.object({
-  messages: z.array(z.object({
-    role: z.enum(['system', 'user', 'assistant']),
+const toolCallSchema = z.object({
+  id: z.string().min(1),
+  type: z.literal('function'),
+  function: z.object({
+    name: z.string().min(1),
+    arguments: z.string(),
+  }),
+});
+
+const systemMessageSchema = z.object({
+  role: z.literal('system'),
+  content: z.string(),
+  name: z.string().optional(),
+});
+
+const userMessageSchema = z.object({
+  role: z.literal('user'),
   content: z.string(),
-  })).min(1),
+  name: z.string().optional(),
+});
+
+const assistantMessageSchema = z.object({
+  role: z.literal('assistant'),
+  content: z.string().nullable().optional(),
+  name: z.string().optional(),
+  tool_calls: z.array(toolCallSchema).optional(),
+}).refine((msg) => {
+  const hasContent = typeof msg.content === 'string' && msg.content.length > 0;
+  const hasToolCalls = (msg.tool_calls?.length ?? 0) > 0;
+  return hasContent || hasToolCalls;
+}, {
+  message: 'assistant messages must include non-empty content or tool_calls',
+});
+
+const toolMessageSchema = z.object({
+  role: z.literal('tool'),
+  content: z.string(),
+  tool_call_id: z.string().min(1),
+  name: z.string().optional(),
+});
+
+const toolDefinitionSchema = z.object({
+  type: z.literal('function'),
+  function: z.object({
+    name: z.string().min(1),
+    description: z.string().optional(),
+    parameters: z.record(z.string(), z.unknown()).optional(),
+    strict: z.boolean().optional(),
+  }),
+});
+
+const toolChoiceSchema = z.union([
+  z.enum(['none', 'auto', 'required']),
+  z.object({
+    type: z.literal('function'),
+    function: z.object({
+      name: z.string().min(1),
+    }),
+  }),
+]);
+
+const chatCompletionSchema = z.object({
+  messages: z.array(z.union([
+    systemMessageSchema,
+    userMessageSchema,
+    assistantMessageSchema,
+    toolMessageSchema,
+  ])).min(1),
   model: z.string().optional(),
   temperature: z.number().min(0).max(2).optional(),
   max_tokens: z.number().int().positive().optional(),
   top_p: z.number().min(0).max(1).optional(),
   stream: z.boolean().optional(),
+  tools: z.array(toolDefinitionSchema).optional(),
+  tool_choice: toolChoiceSchema.optional(),
+  parallel_tool_calls: z.boolean().optional(),
 });
 
 function isRetryableError(err: any): boolean {
@@ -124,8 +191,37 @@ proxyRouter.post('/chat/completions', async (req: Request, res: Response) => {
     return;
   }
 
-  const { messages, temperature, max_tokens, top_p, stream } = parsed.data;
-  const estimatedInputTokens = messages.reduce((sum, m) => sum + Math.ceil(m.content.length / 4), 0);
+  const { temperature, max_tokens, top_p, stream, tools, tool_choice, parallel_tool_calls } = parsed.data;
+  const messages: ChatMessage[] = parsed.data.messages.map((m): ChatMessage => {
+    if (m.role === 'assistant') {
+      return {
+        role: 'assistant',
+        content: m.content ?? null,
+        ...(m.name ? { name: m.name } : {}),
+        ...(m.tool_calls ? { tool_calls: m.tool_calls } : {}),
+      };
+    }
+
+    if (m.role === 'tool') {
+      return {
+        role: 'tool',
+        content: m.content,
+        tool_call_id: m.tool_call_id,
+        ...(m.name ? { name: m.name } : {}),
+      };
+    }
+
+    return {
+      role: m.role,
+      content: m.content,
+      ...(m.name ? { name: m.name } : {}),
+    };
+  });
+
+  const estimatedInputTokens = messages.reduce((sum, m) => {
+    if (typeof m.content !== 'string') return sum;
+    return sum + Math.ceil(m.content.length / 4);
+  }, 0);
   const estimatedTotal = estimatedInputTokens + (max_tokens ?? 1000);
 
   // Sticky session: prefer the same model for multi-turn conversations
@@ -170,7 +266,7 @@ proxyRouter.post('/chat/completions', async (req: Request, res: Response) => {
         let totalOutputTokens = 0;
         const gen = route.provider.streamChatCompletion(
           route.apiKey, messages, route.modelId,
-          { temperature, max_tokens, top_p },
+          { temperature, max_tokens, top_p, tools, tool_choice, parallel_tool_calls },
         );
 
         for await (const chunk of gen) {
@@ -190,7 +286,7 @@ proxyRouter.post('/chat/completions', async (req: Request, res: Response) => {
       } else {
         const result = await route.provider.chatCompletion(
           route.apiKey, messages, route.modelId,
-          { temperature, max_tokens, top_p },
+          { temperature, max_tokens, top_p, tools, tool_choice, parallel_tool_calls },
         );
 
         const totalTokens = result.usage?.total_tokens ?? 0;

shared/types.ts

@@ -66,9 +66,46 @@ export interface FallbackEntry {
 
 // ---- OpenAI-Compatible Types ----
 
+export interface ChatToolCallFunction {
+  name: string;
+  arguments: string;
+}
+
+export interface ChatToolCall {
+  id: string;
+  type: 'function';
+  function: ChatToolCallFunction;
+}
+
+export interface ChatToolFunctionDefinition {
+  name: string;
+  description?: string;
+  parameters?: Record<string, unknown>;
+  strict?: boolean;
+}
+
+export interface ChatToolDefinition {
+  type: 'function';
+  function: ChatToolFunctionDefinition;
+}
+
+export type ChatToolChoice =
+  | 'none'
+  | 'auto'
+  | 'required'
+  | {
+    type: 'function';
+    function: {
+      name: string;
+    };
+  };
+
 export interface ChatMessage {
-  role: 'system' | 'user' | 'assistant';
-  content: string;
+  role: 'system' | 'user' | 'assistant' | 'tool';
+  content: string | null;
+  name?: string;
+  tool_call_id?: string;
+  tool_calls?: ChatToolCall[];
 }
 
 export interface ChatCompletionRequest {
@@ -78,6 +115,9 @@ export interface ChatCompletionRequest {
   max_tokens?: number;
   stream?: boolean;
   top_p?: number;
+  tools?: ChatToolDefinition[];
+  tool_choice?: ChatToolChoice;
+  parallel_tool_calls?: boolean;
 }
 
 export interface ChatCompletionChoice {
@@ -112,7 +152,11 @@ export interface ChatCompletionChunk {
   model: string;
   choices: {
     index: number;
-    delta: Partial<ChatMessage>;
+    delta: {
+      role?: 'assistant';
+      content?: string;
+      tool_calls?: ChatToolCall[];
+    };
     finish_reason: string | null;
   }[];
 }