@inference-gateway/sdk 0.6.1 → 0.7.0

package/CHANGELOG.md

@@ -2,6 +2,18 @@
 
 All notable changes to this project will be documented in this file.
 
+## [0.7.0](https://github.com/inference-gateway/typescript-sdk/compare/v0.6.2...v0.7.0) (2025-05-26)
+
+### ✨ Features
+
+* Implement MCP List Tools ([#13](https://github.com/inference-gateway/typescript-sdk/issues/13)) ([5c0a38c](https://github.com/inference-gateway/typescript-sdk/commit/5c0a38cbe825161c9d5dc1e15f59b31217aebb23))
+
+## [0.6.2](https://github.com/inference-gateway/typescript-sdk/compare/v0.6.1...v0.6.2) (2025-04-30)
+
+### ♻️ Improvements
+
+* Process also groq reasoning models properly ([#12](https://github.com/inference-gateway/typescript-sdk/issues/12)) ([51ce3bb](https://github.com/inference-gateway/typescript-sdk/commit/51ce3bbbbdf03947bb7928e8edc413b977ea092a))
+
 ## [0.6.1](https://github.com/inference-gateway/typescript-sdk/compare/v0.6.0...v0.6.1) (2025-04-28)
 
 ### ♻️ Improvements

package/README.md

@@ -58,6 +58,32 @@ try {
 }
 ```
 
+### Listing MCP Tools
+
+To list available Model Context Protocol (MCP) tools (only available when EXPOSE_MCP is enabled):
+
+```typescript
+import { InferenceGatewayClient } from '@inference-gateway/sdk';
+
+const client = new InferenceGatewayClient({
+  baseURL: 'http://localhost:8080/v1',
+});
+
+try {
+  const tools = await client.listTools();
+  console.log('Available MCP tools:', tools.data);
+
+  // Each tool has: name, description, server, and optional input_schema
+  tools.data.forEach((tool) => {
+    console.log(`Tool: ${tool.name}`);
+    console.log(`Description: ${tool.description}`);
+    console.log(`Server: ${tool.server}`);
+  });
+} catch (error) {
+  console.error('Error:', error);
+}
+```
+
 ### Creating Chat Completions
 
 To generate content using a model:

package/dist/src/client.d.ts

@@ -1,4 +1,4 @@
-import type { Provider, SchemaChatCompletionMessageToolCall, SchemaCompletionUsage, SchemaCreateChatCompletionRequest, SchemaCreateChatCompletionResponse, SchemaCreateChatCompletionStreamResponse, SchemaError, SchemaListModelsResponse } from './types/generated';
+import type { Provider, SchemaChatCompletionMessageToolCall, SchemaCompletionUsage, SchemaCreateChatCompletionRequest, SchemaCreateChatCompletionResponse, SchemaCreateChatCompletionStreamResponse, SchemaError, SchemaListModelsResponse, SchemaListToolsResponse } from './types/generated';
 interface ChatCompletionStreamCallbacks {
     onOpen?: () => void;
     onChunk?: (chunk: SchemaCreateChatCompletionStreamResponse) => void;
@@ -37,6 +37,11 @@ export declare class InferenceGatewayClient {
      * Lists the currently available models.
      */
     listModels(provider?: Provider): Promise<SchemaListModelsResponse>;
+    /**
+     * Lists the currently available MCP tools.
+     * Only accessible when EXPOSE_MCP is enabled.
+     */
+    listTools(): Promise<SchemaListToolsResponse>;
     /**
      * Creates a chat completion.
      */

package/dist/src/client.js

@@ -77,6 +77,15 @@ class InferenceGatewayClient {
         }
         return this.request('/models', { method: 'GET' }, query);
     }
+    /**
+     * Lists the currently available MCP tools.
+     * Only accessible when EXPOSE_MCP is enabled.
+     */
+    async listTools() {
+        return this.request('/mcp/tools', {
+            method: 'GET',
+        });
+    }
     /**
      * Creates a chat completion.
      */
@@ -177,6 +186,10 @@ class InferenceGatewayClient {
                             if (reasoning_content !== undefined) {
                                 callbacks.onReasoning?.(reasoning_content);
                             }
+                            const reasoning = chunk.choices[0]?.delta?.reasoning;
+                            if (reasoning !== undefined) {
+                                callbacks.onReasoning?.(reasoning);
+                            }
                             const content = chunk.choices[0]?.delta?.content;
                             if (content) {
                                 callbacks.onContent?.(content);

package/dist/src/types/generated/index.d.ts

@@ -47,6 +47,27 @@ export interface paths {
         patch?: never;
         trace?: never;
     };
+    '/mcp/tools': {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        /**
+         * Lists the currently available MCP tools
+         * @description Lists the currently available MCP tools. Only accessible when EXPOSE_MCP is enabled.
+         *
+         */
+        get: operations['listTools'];
+        put?: never;
+        post?: never;
+        delete?: never;
+        options?: never;
+        head?: never;
+        patch?: never;
+        trace?: never;
+    };
     '/proxy/{provider}/{path}': {
         parameters: {
             query?: never;
@@ -180,8 +201,8 @@ export interface components {
             retry?: number;
         };
         Endpoints: {
-            models?: string;
-            chat?: string;
+            models: string;
+            chat: string;
         };
         Error: {
             error?: string;
@@ -197,17 +218,19 @@ export interface components {
             content: string;
             tool_calls?: components['schemas']['ChatCompletionMessageToolCall'][];
             tool_call_id?: string;
-            reasoning?: string;
+            /** @description The reasoning content of the chunk message. */
             reasoning_content?: string;
+            /** @description The reasoning of the chunk message. Same as reasoning_content. */
+            reasoning?: string;
         };
         /** @description Common model information */
         Model: {
-            id?: string;
-            object?: string;
+            id: string;
+            object: string;
             /** Format: int64 */
-            created?: number;
-            owned_by?: string;
-            served_by?: components['schemas']['Provider'];
+            created: number;
+            owned_by: string;
+            served_by: components['schemas']['Provider'];
         };
         /** @description Response structure for listing models */
         ListModelsResponse: {
@@ -216,6 +239,53 @@ export interface components {
             /** @default [] */
             data: components['schemas']['Model'][];
         };
+        /** @description Response structure for listing MCP tools */
+        ListToolsResponse: {
+            /**
+             * @description Always "list"
+             * @example list
+             */
+            object: string;
+            /**
+             * @description Array of available MCP tools
+             * @default []
+             */
+            data: components['schemas']['MCPTool'][];
+        };
+        /** @description An MCP tool definition */
+        MCPTool: {
+            /**
+             * @description The name of the tool
+             * @example read_file
+             */
+            name: string;
+            /**
+             * @description A description of what the tool does
+             * @example Read content from a file
+             */
+            description: string;
+            /**
+             * @description The MCP server that provides this tool
+             * @example http://mcp-filesystem-server:8083/mcp
+             */
+            server: string;
+            /**
+             * @description JSON schema for the tool's input parameters
+             * @example {
+             *       "type": "object",
+             *       "properties": {
+             *         "file_path": {
+             *           "type": "string",
+             *           "description": "Path to the file to read"
+             *         }
+             *       },
+             *       "required": [
+             *         "file_path"
+             *       ]
+             *     }
+             */
+            input_schema?: Record<string, never>;
+        };
         FunctionObject: {
             /** @description A description of what the function does, used by the model to choose when and how to call the function. */
             description?: string;
@@ -266,11 +336,8 @@ export interface components {
         /** @description Options for streaming response. Only set this when you set `stream: true`.
          *      */
         ChatCompletionStreamOptions: {
-            /**
-             * @description If set, an additional chunk will be streamed before the `data: [DONE]` message. The `usage` field on this chunk shows the token usage statistics for the entire request, and the `choices` field will always be an empty array. All other chunks will also include a `usage` field, but with a null value.
-             *
-             * @default true
-             */
+            /** @description If set, an additional chunk will be streamed before the `data: [DONE]` message. The `usage` field on this chunk shows the token usage statistics for the entire request, and the `choices` field will always be an empty array. All other chunks will also include a `usage` field, but with a null value.
+             *      */
             include_usage: boolean;
         };
         CreateChatCompletionRequest: {
@@ -292,6 +359,10 @@ export interface components {
             /** @description A list of tools the model may call. Currently, only functions are supported as a tool. Use this to provide a list of functions the model may generate JSON inputs for. A max of 128 functions are supported.
              *      */
             tools?: components['schemas']['ChatCompletionTool'][];
+            /** @description The format of the reasoning content. Can be `raw` or `parsed`.
+             *     When specified as raw some reasoning models will output <think /> tags. When specified as parsed the model will output the reasoning under  `reasoning` or `reasoning_content` attribute.
+             *      */
+            reasoning_format?: string;
         };
         /** @description The function that the model called. */
         ChatCompletionMessageToolCallFunction: {
@@ -350,11 +421,13 @@ export interface components {
         /** @description A chat completion delta generated by streamed model responses. */
         ChatCompletionStreamResponseDelta: {
             /** @description The contents of the chunk message. */
-            content?: string;
+            content: string;
             /** @description The reasoning content of the chunk message. */
             reasoning_content?: string;
+            /** @description The reasoning of the chunk message. Same as reasoning_content. */
+            reasoning?: string;
             tool_calls?: components['schemas']['ChatCompletionMessageToolCallChunk'][];
-            role?: components['schemas']['MessageRole'];
+            role: components['schemas']['MessageRole'];
             /** @description The refusal message generated by the model. */
             refusal?: string;
         };
@@ -418,6 +491,10 @@ export interface components {
             /** @description The object type, which is always `chat.completion.chunk`. */
             object: string;
             usage?: components['schemas']['CompletionUsage'];
+            /** @description The format of the reasoning content. Can be `raw` or `parsed`.
+             *     When specified as raw some reasoning models will output <think /> tags. When specified as parsed the model will output the reasoning under reasoning_content.
+             *      */
+            reasoning_format?: string;
         };
         Config: unknown;
     };
@@ -449,6 +526,18 @@ export interface components {
                 'application/json': components['schemas']['Error'];
             };
         };
+        /** @description MCP tools endpoint is not exposed */
+        MCPNotExposed: {
+            headers: {
+                [name: string]: unknown;
+            };
+            content: {
+                /** @example {
+                 *       "error": "MCP tools endpoint is not exposed. Set EXPOSE_MCP=true to enable."
+                 *     } */
+                'application/json': components['schemas']['Error'];
+            };
+        };
         /** @description ProviderResponse depends on the specific provider and endpoint being called
          *     If you decide to use this approach, please follow the provider-specific documentations.
          *      */
@@ -504,6 +593,8 @@ export type SchemaMessageRole = components['schemas']['MessageRole'];
 export type SchemaMessage = components['schemas']['Message'];
 export type SchemaModel = components['schemas']['Model'];
 export type SchemaListModelsResponse = components['schemas']['ListModelsResponse'];
+export type SchemaListToolsResponse = components['schemas']['ListToolsResponse'];
+export type SchemaMcpTool = components['schemas']['MCPTool'];
 export type SchemaFunctionObject = components['schemas']['FunctionObject'];
 export type SchemaChatCompletionTool = components['schemas']['ChatCompletionTool'];
 export type SchemaFunctionParameters = components['schemas']['FunctionParameters'];
@@ -525,6 +616,7 @@ export type SchemaConfig = components['schemas']['Config'];
 export type ResponseBadRequest = components['responses']['BadRequest'];
 export type ResponseUnauthorized = components['responses']['Unauthorized'];
 export type ResponseInternalError = components['responses']['InternalError'];
+export type ResponseMcpNotExposed = components['responses']['MCPNotExposed'];
 export type ResponseProviderResponse = components['responses']['ProviderResponse'];
 export type RequestBodyProviderRequest = components['requestBodies']['ProviderRequest'];
 export type RequestBodyCreateChatCompletionRequest = components['requestBodies']['CreateChatCompletionRequest'];
@@ -582,6 +674,29 @@ export interface operations {
             500: components['responses']['InternalError'];
         };
     };
+    listTools: {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        requestBody?: never;
+        responses: {
+            /** @description Successful response */
+            200: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    'application/json': components['schemas']['ListToolsResponse'];
+                };
+            };
+            401: components['responses']['Unauthorized'];
+            403: components['responses']['MCPNotExposed'];
+            500: components['responses']['InternalError'];
+        };
+    };
     proxyGet: {
         parameters: {
             query?: never;

package/dist/tests/client.test.js

@@ -57,6 +57,7 @@ describe('InferenceGatewayClient', () => {
                         object: 'model',
                         created: 1686935002,
                         owned_by: 'openai',
+                        served_by: generated_1.Provider.openai,
                     },
                 ],
             };
@@ -81,6 +82,53 @@ describe('InferenceGatewayClient', () => {
             await expect(client.listModels(generated_1.Provider.openai)).rejects.toThrow(errorMessage);
         });
     });
+    describe('listTools', () => {
+        it('should fetch available MCP tools', async () => {
+            const mockResponse = {
+                object: 'list',
+                data: [
+                    {
+                        name: 'read_file',
+                        description: 'Read content from a file',
+                        server: 'http://mcp-filesystem-server:8083/mcp',
+                    },
+                    {
+                        name: 'write_file',
+                        description: 'Write content to a file',
+                        server: 'http://mcp-filesystem-server:8083/mcp',
+                    },
+                ],
+            };
+            mockFetch.mockResolvedValueOnce({
+                ok: true,
+                json: () => Promise.resolve(mockResponse),
+            });
+            const result = await client.listTools();
+            expect(result).toEqual(mockResponse);
+            expect(mockFetch).toHaveBeenCalledWith('http://localhost:8080/v1/mcp/tools', expect.objectContaining({
+                method: 'GET',
+                headers: expect.any(Headers),
+            }));
+        });
+        it('should throw error when MCP is not exposed', async () => {
+            const errorMessage = 'MCP not exposed';
+            mockFetch.mockResolvedValueOnce({
+                ok: false,
+                status: 403,
+                json: () => Promise.resolve({ error: errorMessage }),
+            });
+            await expect(client.listTools()).rejects.toThrow(errorMessage);
+        });
+        it('should throw error when unauthorized', async () => {
+            const errorMessage = 'Unauthorized';
+            mockFetch.mockResolvedValueOnce({
+                ok: false,
+                status: 401,
+                json: () => Promise.resolve({ error: errorMessage }),
+            });
+            await expect(client.listTools()).rejects.toThrow(errorMessage);
+        });
+    });
     describe('createChatCompletion', () => {
         it('should create a chat completion', async () => {
             const mockRequest = {
@@ -391,6 +439,164 @@ describe('InferenceGatewayClient', () => {
                 }),
             }));
         });
+        it('should handle streaming chat completions with reasoning field', async () => {
+            const mockRequest = {
+                model: 'groq/deepseek-distilled-llama-3.1-70b',
+                messages: [{ role: generated_1.MessageRole.user, content: 'Hello' }],
+            };
+            const mockStream = new web_1.TransformStream();
+            const writer = mockStream.writable.getWriter();
+            const encoder = new node_util_1.TextEncoder();
+            mockFetch.mockResolvedValueOnce({
+                ok: true,
+                body: mockStream.readable,
+            });
+            const callbacks = {
+                onOpen: jest.fn(),
+                onChunk: jest.fn(),
+                onReasoning: jest.fn(),
+                onContent: jest.fn(),
+                onFinish: jest.fn(),
+            };
+            const streamPromise = client.streamChatCompletion(mockRequest, callbacks);
+            await writer.write(encoder.encode('data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"groq/deepseek-distilled-llama-3.1-70b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"groq/deepseek-distilled-llama-3.1-70b","choices":[{"index":0,"delta":{"content":"","reasoning":"Let me"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"groq/deepseek-distilled-llama-3.1-70b","choices":[{"index":0,"delta":{"content":"","reasoning":" think"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"groq/deepseek-distilled-llama-3.1-70b","choices":[{"index":0,"delta":{"content":"","reasoning":" about"},"finish_reason":"stop"}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"groq/deepseek-distilled-llama-3.1-70b","choices":[{"index":0,"delta":{"content":"","reasoning":" this"},"finish_reason":"stop"}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"groq/deepseek-distilled-llama-3.1-70b","choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"groq/deepseek-distilled-llama-3.1-70b","choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}]}\n\n' +
+                'data: [DONE]\n\n'));
+            await writer.close();
+            await streamPromise;
+            expect(callbacks.onOpen).toHaveBeenCalledTimes(1);
+            expect(callbacks.onChunk).toHaveBeenCalledTimes(7);
+            expect(callbacks.onReasoning).toHaveBeenCalledTimes(4);
+            expect(callbacks.onReasoning).toHaveBeenCalledWith('Let me');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' think');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' about');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' this');
+            expect(callbacks.onContent).toHaveBeenCalledTimes(2);
+            expect(callbacks.onContent).toHaveBeenCalledWith('Hello');
+            expect(callbacks.onContent).toHaveBeenCalledWith('!');
+            expect(callbacks.onFinish).toHaveBeenCalledTimes(1);
+            expect(mockFetch).toHaveBeenCalledWith('http://localhost:8080/v1/chat/completions', expect.objectContaining({
+                method: 'POST',
+                body: JSON.stringify({
+                    ...mockRequest,
+                    stream: true,
+                    stream_options: {
+                        include_usage: true,
+                    },
+                }),
+            }));
+        });
+        it('should handle streaming chat completions with reasoning_content (DeepSeek)', async () => {
+            const mockRequest = {
+                model: 'deepseek/deepseek-reasoner',
+                messages: [{ role: generated_1.MessageRole.user, content: 'Hello' }],
+            };
+            const mockStream = new web_1.TransformStream();
+            const writer = mockStream.writable.getWriter();
+            const encoder = new node_util_1.TextEncoder();
+            mockFetch.mockResolvedValueOnce({
+                ok: true,
+                body: mockStream.readable,
+            });
+            const callbacks = {
+                onOpen: jest.fn(),
+                onChunk: jest.fn(),
+                onReasoning: jest.fn(),
+                onContent: jest.fn(),
+                onFinish: jest.fn(),
+            };
+            const streamPromise = client.streamChatCompletion(mockRequest, callbacks);
+            await writer.write(encoder.encode('data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"content":"","reasoning_content":"This"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"content":"","reasoning_content":" is"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"content":"","reasoning_content":" a"},"finish_reason":"stop"}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"content":"","reasoning_content":" reasoning"},"finish_reason":"stop"}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"content":"","reasoning_content":" content"},"finish_reason":"stop"}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"deepseek/deepseek-reasoner","choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}]}\n\n' +
+                'data: [DONE]\n\n'));
+            await writer.close();
+            await streamPromise;
+            expect(callbacks.onOpen).toHaveBeenCalledTimes(1);
+            expect(callbacks.onChunk).toHaveBeenCalledTimes(8);
+            expect(callbacks.onReasoning).toHaveBeenCalledTimes(5);
+            expect(callbacks.onReasoning).toHaveBeenCalledWith('This');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' is');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' a');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' reasoning');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' content');
+            expect(callbacks.onContent).toHaveBeenCalledTimes(2);
+            expect(callbacks.onContent).toHaveBeenCalledWith('Hello');
+            expect(callbacks.onContent).toHaveBeenCalledWith('!');
+            expect(callbacks.onFinish).toHaveBeenCalledTimes(1);
+            expect(mockFetch).toHaveBeenCalledWith('http://localhost:8080/v1/chat/completions', expect.objectContaining({
+                method: 'POST',
+                body: JSON.stringify({
+                    ...mockRequest,
+                    stream: true,
+                    stream_options: {
+                        include_usage: true,
+                    },
+                }),
+            }));
+        });
+        it('should handle streaming chat completions with reasoning field (Groq)', async () => {
+            const mockRequest = {
+                model: 'llama-3.1-70b-versatile',
+                messages: [{ role: generated_1.MessageRole.user, content: 'Hello' }],
+            };
+            const mockStream = new web_1.TransformStream();
+            const writer = mockStream.writable.getWriter();
+            const encoder = new node_util_1.TextEncoder();
+            mockFetch.mockResolvedValueOnce({
+                ok: true,
+                body: mockStream.readable,
+            });
+            const callbacks = {
+                onOpen: jest.fn(),
+                onChunk: jest.fn(),
+                onReasoning: jest.fn(),
+                onContent: jest.fn(),
+                onFinish: jest.fn(),
+            };
+            const streamPromise = client.streamChatCompletion(mockRequest, callbacks);
+            await writer.write(encoder.encode('data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"llama-3.1-70b-versatile","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"llama-3.1-70b-versatile","choices":[{"index":0,"delta":{"content":"","reasoning":"Let me"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"llama-3.1-70b-versatile","choices":[{"index":0,"delta":{"content":"","reasoning":" think"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"llama-3.1-70b-versatile","choices":[{"index":0,"delta":{"content":"","reasoning":" about"},"finish_reason":"stop"}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"llama-3.1-70b-versatile","choices":[{"index":0,"delta":{"content":"","reasoning":" this"},"finish_reason":"stop"}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"llama-3.1-70b-versatile","choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]}\n\n' +
+                'data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677652288,"model":"llama-3.1-70b-versatile","choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}]}\n\n' +
+                'data: [DONE]\n\n'));
+            await writer.close();
+            await streamPromise;
+            expect(callbacks.onOpen).toHaveBeenCalledTimes(1);
+            expect(callbacks.onChunk).toHaveBeenCalledTimes(7);
+            expect(callbacks.onReasoning).toHaveBeenCalledTimes(4);
+            expect(callbacks.onReasoning).toHaveBeenCalledWith('Let me');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' think');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' about');
+            expect(callbacks.onReasoning).toHaveBeenCalledWith(' this');
+            expect(callbacks.onContent).toHaveBeenCalledTimes(2);
+            expect(callbacks.onContent).toHaveBeenCalledWith('Hello');
+            expect(callbacks.onContent).toHaveBeenCalledWith('!');
+            expect(callbacks.onFinish).toHaveBeenCalledTimes(1);
+            expect(mockFetch).toHaveBeenCalledWith('http://localhost:8080/v1/chat/completions', expect.objectContaining({
+                method: 'POST',
+                body: JSON.stringify({
+                    ...mockRequest,
+                    stream: true,
+                    stream_options: {
+                        include_usage: true,
+                    },
+                }),
+            }));
+        });
     });
     describe('proxy', () => {
         it('should proxy requests to a specific provider', async () => {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inference-gateway/sdk",
-  "version": "0.6.1",
+  "version": "0.7.0",
   "description": "An SDK written in Typescript for the [Inference Gateway](https://github.com/inference-gateway/inference-gateway).",
   "main": "dist/src/index.js",
   "types": "dist/src/index.d.ts",