Gpt-realtime: Where is the documentation for image input and MCP?

Documentation
1

In the blog post for gpt-realtime there are primitive examples for how to incorporate image input and MCP. For each example, the blog post references the official documentation for more information.

As of September 3, 2025 the links provided do not point anywhere and instead resolve to other links in the old documentation:

MCP:
http://platform.openai.com/docs/guides/realtime-mcp resolves to https://platform.openai.com/docs/guides/realtime-conversations

Image input:
http://platform.openai.com/docs/guides/realtime-inputs-outputs resolves to https://platform.openai.com/docs/guides/realtime-conversations

Has anyone seen this new documentation, or know when it’s coming?

5 Likes
2

I have the same question—I also want to know how to use an MCP server with the Realtime API.

3

We noticed that the realtime AI docs were updated when the new version came out, but there’s loads of mistakes and omissions.

2 Likes