OpenAI has significantly upgraded its Responses API introducing powerful tools like native image generation and Code Interpreter. The update also adds crucial support for remote Model Context Protocol (MCP) servers, aiming to simplify building advanced AI agents. These enhancements are designed for developers and enterprises seeking more capable, action-oriented applications.
This evolution builds on the API’s initial March 2025 introduction, part of a strategy to provide a unified toolkit for third-party developers. OpenAI stated then its Responses API “combines the simplicity of Chat Completions with the tool-use capabilities of the Assistants API.” The Assistants API is planned for a mid-2026 phase-out.
Since its debut, hundreds of thousands of developers have utilized the API, processing trillions of tokens for diverse applications, as detailed by OpenAI.
Key Tooling And Interoperability Enhancements
A central feature of the May 21 update is support for remote Model Context Protocol (MCP) servers. MCP standardizes how applications provide context to large language models, and could become a universal language for enterprise AI.
This allows OpenAI’s models to connect with external tools from providers like Cloudflare, HubSpot, and Stripe using minimal code. Reinforcing this direction, OpenAI has joined the MCP steering committee.
In order to best support the ecosystem and contribute to this developing standard, OpenAI has also joined the steering committee for MCP.
— OpenAI Developers (@OpenAIDevs) May 21, 2025
Get started with MCP <> Responses API: https://t.co/VkNwtKm4ta
Furthering this interoperability, Zapier announced its MCP can now be used with the Responses API. Reid Robinson from Zapier Blog explained “This new powerful connection with OpenAI’s Responses API enables GPT-4.1 and other supported models to directly interact with Zapier’s ecosystem of nearly 8,000 app integrations,” facilitating tasks such as sending Slack messages or updating CRM records.
The API now integrates `gpt-image-1`, a variant of OpenAI’s GPT-4o image generation model. This tool, supported on the o3 reasoning model, offers real-time streaming previews and multi-turn editing. Alongside this, the Code Interpreter tool allows models to handle data analysis, complex math, and logic-based tasks. File search capabilities were also upgraded, enabling searches across multiple vector stores and attribute-based filtering.
Enterprise Focus And Developer Experience
OpenAI has rolled out several features targeting enterprise needs and improving the developer experience. A background mode handles long-running asynchronous tasks, addressing potential timeouts. For enhanced transparency, reasoning summaries provide natural-language explanations of a model’s thought process at no additional cost. Encrypted reasoning items offer an added privacy layer for Zero Data Retention customers.
These updates are supported across OpenAI’s GPT-4o series, GPT-4.1 series, and o-series reasoning models. Notably, the o3 and o4-mini models can now call tools and functions directly within their chain-of-thought in the Responses API. This preserves reasoning tokens across requests, improving model intelligence and reducing cost and latency, according to OpenAI. Early adopters of the Responses API, such as Zencoder, Revi, and MagicSchool AI, have already been building varied agentic applications.
Pricing, Availability, And Strategic Context
Despite the expanded feature set, OpenAI confirmed pricing for these new tools and capabilities remains consistent with existing rates. For example, Code Interpreter costs $0.03 per session, while image generation via `gpt-image-1` starts at $0.011 per image. The Azure OpenAI service also offers the Responses API in preview, as detailed by Microsoft Learn, though with some initial tool limitations compared to OpenAI’s direct offering.
The Responses API is positioned as OpenAI’s long-term direction for agentic applications, designed for multimedia input and intended to replace the Assistants API beta. This ongoing development reflects OpenAI’s broader strategy to equip developers with sophisticated AI tools, even as the industry navigates security and regulatory considerations for increasingly autonomous systems.
