OpenAI has released a new suite of tools designed to help developers build AI agents similar to the company’s own Deep Research and Operator. The new offerings include the Responses API and the open-source Agents SDK, which provide developers with the building blocks to create AI applications that can search the web, analyze files, and even control computer interfaces.
Responses API
The Responses API combines features of OpenAI’s Chat Completions API with tool-use functionality, allowing developers to leverage multiple built-in tools within a single API call. This new API comes with three key tools:
- Web Search: Provides real-time information with citations, powered by the same technology behind ChatGPT’s search capability
- File Search: Enables AI to retrieve relevant information from large document collections, supporting multiple file formats
- Computer Use Tool: Allows AI to interact with computer interfaces, performing tasks like data entry and web navigation
According to Olivier Godement, head of product for OpenAI’s platform, the company recognizes it can’t build every possible AI agent itself. “We’re super excited to provide those foundations, those building blocks for developers to build the best agents for their use case, their needs,” he told The Verge.
Agents SDK
In addition to the Responses API, OpenAI has released the Agents SDK, an open-source toolkit that helps developers manage and coordinate agent workflows. Notably, the SDK supports not just OpenAI models but also those from competitors like Anthropic, Google, and open-source options from DeepSeek, Qwen, Mistral, and Meta’s Llama family.
“The Responses API is like this atomic unit of using models and tools to do a particular thing,” explained Nikunj Handa, a product manager for OpenAI’s API team. “The Agents SDK is having multiple of those atomic units work together to solve even more complicated tasks.”
Ready for action
OpenAI has positioned these tools as enterprise-ready solutions that can be integrated into existing systems without extensive custom development. The pricing structure varies by tool, with web search included in standard API usage, file search at $2.50 per thousand queries (plus storage fees), and the computer use tool at $3 per million input tokens and $12 per million output tokens.
The company plans to replace its Assistants API with the Responses API by mid-2026, incorporating improvements based on developer feedback. Both new offerings are available to developers starting today, with documentation and an API playground available on OpenAI’s website.
Sources: VentureBeat, The Verge