Runway Brings Video Generation to ChatGPT via MCP

A new Model Context Protocol integration lets GPT-6 Astra handle scripting, scene direction, and video rendering in a single conversation thread.

Runway announced the release of Runway MCP on October 7, 2026. The connector relies on Anthropic’s open Model Context Protocol to bring external video generation tools directly into ChatGPT. According to technical documentation published by the company, paid subscribers can now draft prompts, refine scene logic, and trigger full video renders without leaving their chat window.

Initial rollout focuses on OpenAI’s GPT-6 Astra model, turning the chat interface into an orchestration layer for video generation.

How the chat workflow operates

The setup removes the need to juggle browser tabs or copy manual API keys between dashboards. Inside ChatGPT, users mention @Runway and provide a creative brief, still images, or storyboard references. GPT-6 Astra organizes the scene composition, sends instructions to Runway’s infrastructure, and embeds the finished video clip directly in the chat feed.

Account authentication runs over standard OAuth linked to the user’s Runway profile. While installing the connector requires no setup fee, generating clips debits standard plan credits from the user’s active Runway subscription. Rendered files synchronize automatically with the user’s cloud library on Runway.

Multiple video models under one interface

The integration is not limited to a single pipeline. It gives users access to Runway’s proprietary foundation models, including Gen-4.5, Gen-4 Turbo, and Gen-4 Image. The connector also surfaces third-party video and image models distributed across Runway’s catalog, such as Kling 3.0, Veo 3.1, Seedance 2.5, GPT Image 2, and Nano Banana Pro.

Because the service uses the open MCP standard, Runway is deploying the same server endpoint outside OpenAI’s ecosystem. The server connects to Claude, Cursor, and Replit, allowing production teams to call rendering pipelines directly from coding environments and standalone AI workspaces.

Filling the post-Sora gap in ChatGPT

The release fills a practical gap left inside ChatGPT after OpenAI discontinued Sora earlier in 2026. While the chat platform retained native tools for text and image creation, it lacked an integrated video engine. By adopting MCP, OpenAI delegates media synthesis to specialized providers while using GPT-6 Astra to manage the prompt workflow.

Access requires a paid ChatGPT subscription, covering Plus, Pro, Business, and Enterprise plans. Free accounts are excluded. For teams running automated workflows, questions remain: while standard jobs pull from regular Runway credits, the company has not yet clarified whether dedicated pricing tiers or enterprise platform fees will apply to high-volume MCP usage.