OpenAI rolls out MCP-aligned Apps and Agents SDK
OpenAI has begun rolling out Model Context Protocol (MCP) support across its ChatGPT platform, making standardized tool connectivity available to enterprise and education customers as of July 2026.
MCP Support Now in Beta
Full MCP support, including modify and write actions, is rolling out in beta for ChatGPT Business, Enterprise, and Education plans OpenAI Help. Organizations can now build, test, and deploy MCP-powered applications directly within ChatGPT's developer mode, streamlining the process of connecting external tools and data sources to agent workflows.
The rollout aligns OpenAI's platform with the Model Context Protocol standard, a specification designed to enable consistent tool integration across AI systems. This approach reduces fragmentation and allows developers to write once and deploy across multiple agent contexts.
Developer Access and Capabilities
Pro users retain access to the Apps SDK for building custom applications OpenAI Help, while the MCP integration deepens capabilities for organizations operating at scale. The beta rollout prioritizes Business, Enterprise, and Education tiers, suggesting OpenAI is targeting organizations with mission-critical agent deployments and compliance requirements.
Developers can now leverage MCP's standardized protocol to connect agents to databases, APIs, and internal systems without custom middleware layers. The modify and write action support indicates agents can execute state-changing operations—not merely read data—through connected tools.
Why This Matters for Agent Development
MCP-style tool connectivity removes a structural barrier in agent development: the need to build bespoke integrations for each new tool or data source. By making the protocol a first-class part of ChatGPT's stack, OpenAI reduces the operational overhead of deploying agents in production environments.
This is particularly significant for enterprises running multiple agent instances across departments. Standardized tool connectivity means security teams can audit and control tool access through a single framework, rather than managing custom integration code for each agent deployment.
The beta status suggests the feature is still maturing—enterprises should expect iterative improvements and potential breaking changes as OpenAI refines the MCP implementation. Documentation and migration guidance for teams already building custom integrations has begun circulating among developer communities OpenAI Help.
No pricing changes or additional fees have been announced for MCP support within existing ChatGPT plan tiers.