Text Generation (Responses)

Model input supports the following formats: string: plain text, such as "Hello"; array: message array, arranged in conversational order.

An array of tools that the model can call when generating responses. Supports both built-in tools and custom function tools, which can be used in combination. For optimal response quality, it is recommended to enable the code_interpreter, web_search, and web_extractor tools simultaneously. web_search: A web search tool that allows the model to search for the latest information on the internet. Related documentation: Web Search. Properties: type string (Required). Fixed as web_search. Usage example: [{"type": "web_search"}]. web_extractor: A web extraction tool that allows the model to access and extract web page content. Currently, it must be used in conjunction with the web_search tool. qwen3-max and qwen3-max-2026-01-23 require thinking mode to be enabled simultaneously. Related documentation: Web Extraction. Properties: code_interpreter: A code interpreter tool that allows the model to execute code and return results, supporting data analysis. qwen3-max and qwen3-max-2026-01-23 require thinking mode to be enabled simultaneously. Related documentation: Code Interpreter. Properties: web_search_image: Searches for images based on text descriptions. Related documentation: Text-to-Image Search. Properties: type string (Required). Fixed as web_search_image. Usage example: [{"type": "web_search_image"}]. image_search: Searches for similar or related images based on an image; the input must include the image URL. Related documentation: Image-to-Image Search. Properties: type string (Required). Fixed as image_search. Usage example: [{"type": "image_search"}]. file_search: Searches within uploaded or associated knowledge bases. Related documentation: Knowledge Retrieval. Properties: type string (Required). Fixed as file_search. vector_store_ids array (Required). The ID of the knowledge base to retrieve. Currently, only one knowledge base ID is supported. Usage example: [{"type": "file_search", "vector_store_ids": ["your_knowledge_base_id"]}]. MCP Call: Calls external services via MCP (Model Context Protocol). Related documentation: MCP. Properties: type string (Required). Fixed as mcp. server_protocol string (Required). The communication protocol with the MCP service, such as "sse". server_label string (Required). The service label used to identify the MCP service. server_description string (Optional). The service description to help the model understand its functions and applicable scenarios. server_url string (Required). The URL of the MCP service endpoint. headers object (Optional). Request headers for carrying authentication and other information, such as Authorization. Usage example: mcp_tool = { "type": "mcp", "server_protocol": "sse", "server_label": "amap-maps", "server_description": "The Amap Maps MCP Server now covers 15 core interfaces, providing full-scenario geographic information services, including generating custom maps, navigating to destinations, hailing rides, geocoding, reverse geocoding, IP positioning, weather queries, cycling route planning, walking route planning, driving route planning, transit route planning, distance measurement, keyword search, nearby search, and detail search.", "server_url": "https://dashscope.aliyuncs.com/api/v1/mcps/amap-maps/sse", "headers": { "Authorization": "Bearer <your-mcp-server-token>" } } Custom function tool: A custom function tool that allows the model to call functions you define. When the model determines that a tool call is needed, the response will return an output of type function_call. Related documentation: Function Calling. Properties: type string (Required). Must be set to function. name string (Required). The tool name. Only letters, numbers, underscores (_), and hyphens (-) are allowed, up to 64 Tokens. description string (Required). The tool description to help the model determine when and how to call the tool. parameters object (Optional). The parameter description of the tool, which must be a valid JSON Schema. If the parameters argument is empty, it indicates that the tool has no input parameters (e.g., a time query tool). To improve the accuracy of tool calls, it is recommended to pass parameters. Usage example: [{ "type": "function", "name": "get_weather", "description": "Get weather information for a specified city", "parameters": { "type": "object", "properties": { "city": { "type": "string", "description": "City name" } }, "required": ["city"] } }]

Controls how the model selects and calls tools. This parameter supports two assignment formats: string mode and object mode.

Control the model's reasoning effort. The model reasons before responding, and the reasoning content will be returned via output items of the reasoning type.