OpenAI has introduced new tools and APIs to support the development of AI agents, launching the Responses API (Application Programming Interface) on Tuesday.
The new API integrates features from the Chat Completions and Assistants APIs, allowing developers to build AI agents that perform tasks using multiple tools in a single call. The company also released built-in tools, including web search, file search, and computer use, along with an open-source Agents SDK for multi-agent workflows.
The Responses API is designed to simplify AI agent development by providing a structured approach to tool use, data storage, and evaluation.
Web search in the API provides information with citations and is available with GPT-4o and GPT-4o-mini. File search retrieves relevant information from documents, supporting query optimization and metadata filtering.
The computer use tool, available in the research preview, enables the automation of browser-based and desktop workflows by capturing mouse and keyboard actions.
The launch follows increased activity from Chinese AI startups introducing new models. Chinese startup Monica has introduced its AI agent Manus, while DeepSeek has gained attention in the U.S. technology sector.
Monica claims Manus outperforms OpenAI’s DeepResearch agent and has announced a partnership with the team behind Alibaba’s Qwen AI models.
OpenAI has stated that the Responses API will replace the Assistants API, which is set to be phased out by mid-2026. The company plans to release additional tools in the coming months to develop AI agent applications on its platform further.
