Runway makes its generation models accessible via Claude, ChatGPT, or Cursor

Runway opens its Gen-4.5 and Gen-4 Turbo models to MCP-compatible agents like Claude, ChatGPT, Cursor, and Replit to generate video directly in workflows.

Runway is opening its image and video generation models to MCP-compatible agents, including Claude, ChatGPT, Cursor, and Replit, and more broadly to any tool supporting the protocol. The agent calls the platform in the background and returns the output to the conversation, without needing to access the interface: a product URL yields a launch video, a reference image yields a multi-shot spot with dialogue, all remaining within the current workflow. Behind the connector, the catalog aggregates the in-house range, Gen-4.5 and Gen-4 Turbo, and third-party models like Seedance 2.0, Kling, GPT Image 2, Veo, or Nano Banana Pro, with the agent making the decision alone or as instructed, within the limits of the subscription. Billing remains unchanged: standard Runway credits, generations archived in the library, parity between solo and team usage on the same infrastructure as the application.