Runway releases Agent 2.0: Video generation company turns to "creative AI platform"
Runway released Agent 2.0 on June 25. This is a major upgrade of its creative AI Agent product line. Users describe their needs in natural language, and the Agent generates, edits and iterates cross-modally among videos, images, and audios; marking Runway's strategic transformation from a "video generation model company" to a "creative AI platform".
Agent 2.0, released by Runway on June 25, is more of a transformation manifesto than a product upgrade. This company, which started with the Gen series video model, is redefining itself as a "creative AI platform" - Agent 2.0 is the core entrance of this transformation: users describe creative needs in natural language, and Agent completes generation, editing and iteration among video, image, audio, and text.
Dual-line promotion of models and products
Spreading out Runway's actions in 2026, a two-line path of "model + product" is clearly visible:
- Model layer: General World Models (GWM-1), Gen-4.5, Aleph 2.0 and video generation models
- Product Tier: Agent 2.0 (June 25), Runway Media Router and Runway Dev (July 24), Aleph 2.0 coming to Figma Weave (June 22)
Agent 2.0 is at the core of this platforming strategy—it’s the glue that connects model capabilities to creative workflows.
Signal meaning of "cross-modal Agent"
There is a sentence in the official description that is worth chewing over and over: "agents research, generate, and refine across video, image, audio, and text". This reflects a fundamental shift in the AI video track—from "single model generation" to "multi-modal agent workflow." Creators no longer call video models, image models, and audio models one by one, but let an Agent connect all modalities to complete the entire creative link.
From an industry perspective, Runway's move has raised the competitive dimension from "whose model is better" to "whose platform is complete": Luma (Skills + Ray), Keling, and Pika are all working in the same direction. For domestic AI video manufacturers, Agent 2.0 reminds us of a reality - when model capabilities converge, the winner is turning to "whether it can provide an end-to-end workflow from conception to production" and "whether Agent can be embedded in the daily tools of professional creative teams."
Several directions worth tracking in the future:
- Cross-modal consistency of Agent 2.0: The unity of characters, styles, audio and video when multi-modal is connected is the key to the success or failure of the experience.
- Collaboration with Media Router: Agent scheduling and generation tasks + closed-loop maturity of routing management assets.
- Adoption by professional creative teams: Whether advertising, film and television companies regard it as a production tool rather than a toy.
- Agentization of domestic AI video platforms: the follow-up speed of workflow products such as Keling and Jimeng.
Reviews