As of May 21, 2025, we’ve added new built-in tools to the Responses API—remote MCP servers, Image Generation, Code Interpreter, and an upgraded File Search—along with background mode and encrypted content, so you can build agents that pull richer context and run more reliably. For details, see our Responses API docs.
The extent to which OpenAI API models have knowledge of current events depends on the specific model. Our models overview page contains a detailed overview of the knowledge cutoffs for each model.
Provide models with current information
To bridge the gap between a model's training cutoff and current events, you can use built-in search tools available through OpenAI's Responses API, such as:
Web Search: Allows the model to access up-to-date information directly from the internet.
File Search: Enables the model to retrieve information from files you upload to vector stores.
These tools can retrieve information beyond the model's training data. Web search can provide current information from the web; file search retrieves information from the files you provide.
Enable access to current information
To equip your API models with current knowledge:
Integrate the Responses API in your workflow.
Activate the Web Search and/or File Search tools when making API calls.
The model decides whether to use an enabled tool based on your prompt. You can control tool use with the
tool_choiceparameter.
Learn more about using the Responses API and Web and File Search.