OpenAI releases a big move! Core API supports MCP, changing the development of intelligent agents overnight
OpenAI Responses API supports MCP, which brings revolutionary breakthrough to intelligent agent development.
Core content:
1. Responses API supports MCP, simplifying the connection between intelligent agents and external services
2. Developers can achieve efficient integration of intelligent agents and external tools with just a few lines of code
3. MCP centralizes management tools to improve intelligent agent performance and enhance security
This morning, the news that OpenAI acquired io occupied most of the headlines. At the same time, OpenAI also "quietly" released another blockbuster news that the core API for developing intelligent agents - Responses API supports MCP services.
In the traditional method, when developing intelligent agents, we need to interact with external services through function calls. Each operation involves network transmission from the large model to the backend and then to the external service, resulting in multiple jumps, high latency, and increased complexity in expansion and management.
Responses APIOther new features
In addition to supporting MCP , OpenAI has also made major updates to the image generation, code interpreter , and file search tools in the Responses API , further enhancing the capabilities of the agents.
Image Generation: Developers can now directly access OpenAI ’s latest image generation models (such as <gpt-image-1> ) in the Responses API and use them as tools. The tool supports live streaming, allowing developers to see a preview of the image generation process, and supports multiple rounds of editing, allowing developers to gradually fine-tune the image.
Code Interpreter : The Code Interpreter tool is now available in the Responses API , supporting data analysis, solving complex math and coding problems, and even helping models deeply understand and manipulate images. For example, when working with math problems, models can use Code Interpreter to run code to get the answer, significantly improving performance.
File Search: The file search tool has been enhanced to allow developers to extract relevant content blocks from documents into the context of the model based on user queries. In addition, the tool supports performing searches across multiple vector stores and allows attribute filtering using arrays.
At the same time, OpenAI also introduced new features in the Responses API .
Background mode: For tasks that take a long time to process, developers can use background mode to start these tasks asynchronously without worrying about timeouts or other connection issues. Developers can poll these tasks to check if they are completed, or start streaming events when needed.
Reasoning summaries: The Responses API can now generate concise natural language summaries of the model's internal thought chain. This makes it easier for developers to debug, audit, and build better end-user experiences.
Encrypted Inference Items: Zero Data Retention ( ZDR ) eligible customers can reuse inference items between API requests without storing any inference items on OpenAI ’s servers. This not only improves intelligence, but also reduces labeling usage, lowering costs and latency.
