OpenAI adopts Anthropic's open MCP standard

OpenAI just decided that when your biggest rival invents a universal plug, the smartest move is to simply plug in.

On Wednesday, OpenAI CEO Sam Altman announced that the company will roll out support for Anthropic’s open-source Model Context Protocol across its lineup of products, starting with its Agents SDK and expanding to the ChatGPT desktop app and Responses API soon.

A rare truce in the AI standards race

Altman took to X to share the news, noting that people love MCP and that OpenAI is excited to add support across its products. While the Agents SDK gets MCP right out of the gate, Altman confirmed that ChatGPT desktop users and developers using the Responses API will see integrations roll out shortly. OpenAI added that it will detail more of its MCP roadmap over the coming months.

Anthropic was quick to accept the nod. Anthropic Chief Product Officer Mike Krieger posted on X to welcome the news, noting that MCP has already grown into a thriving open standard with thousands of integrations. In Krieger’s words, large language models are at their best when they connect directly to the software and data people already rely on daily.

How Anthropic’s protocol won over devs

For anyone building AI tooling, connecting models to internal company tools, private databases, and dev environments used to mean writing endless custom glue code.

MCP solves that by creating a standardized two-way bridge between models and data. Developers set up MCP servers that expose specific systems, and MCP clients—like chatbots or automated agents—can query those servers whenever a task calls for it.

Since Anthropic open-sourced the protocol a few months ago, adoption has moved fast. Tech companies including Block, Apollo, Replit, Codeium, and Sourcegraph have already rolled out MCP integrations across their own platforms. OpenAI’s adoption means that developers no longer have to pick sides when building custom integrations.

Why it matters

We think this is one of the most pragmatic moves OpenAI has made this year. Rather than wasting engineering bandwidth and fragmenting the developer community with a proprietary connector, OpenAI recognized that Anthropic had already won developer hearts with MCP. Swallowing some pride instantly hands OpenAI access to thousands of pre-built integrations.

For developers and regular users, the payoff is immediate. If you build an MCP server for your internal tools, your codebase, or your database, you can now hook it up to Claude and ChatGPT without duplicating your work. It lowers the barrier to building useful agents and makes swapping underlying models dramatically easier.

Across the industry, this effectively crowns MCP as the USB-C of the AI ecosystem. When two fiercest competitors rally around the exact same open-source protocol, the debate over how models should talk to external data is pretty much settled.