Runway adds Seedance 2.0, GPT Image 2, and Nano Banana Pro to its API

• Runway said its API now bundles multiple outside and in-house generation systems in one place, including Seedance 2.0, GPT Image 2, HappyHorse 1.0, Nano Banana Pro, and Magnific Precision Upscaler V2, according to runwayml's API announcement.
• The API pitch is less about one flagship model than a cross-model stack: the same developer portal now points creators toward image generation, video generation, and upscaling from a single surface, per runwayml's developer link post.
• Early creator posts already show that stack in practice, with juliewdesign_'s clip combining Midjourney, Nano Banana Pro, and Seedance, while AllaAisling's post pairs Seedance 2.0 with a separate 4K upscale step.
• Runway also shipped an MCP connector two days earlier, and runwayml's MCP launch post says it plugs the same model catalog into Claude, ChatGPT, Cursor, Replit, and other agent surfaces.

Runway's useful reveal here is packaging. You can jump from the main developer portal into a menu that spans Seedance 2.0, GPT Image 2, Nano Banana Pro, and Magnific's upscaler, while the separate Runway MCP page extends that catalog into Claude, ChatGPT, Cursor, and Replit. The creator examples are already hybrid, with juliewdesign_'s post mixing three tools in one piece instead of treating any single model as the whole workflow.

Runway framed the update as an aggregation move: "all of the models you need, all in one place," with Seedance 2.0, GPT Image 2, HappyHorse 1.0, Nano Banana Pro, and Magnific Precision Upscaler V2 named in the launch post.

That gives the API a broader creative menu than a normal single-model release. The named additions break into three workflow buckets:

• Image and asset generation: Seedance 2.0, GPT Image 2, Nano Banana Pro, HappyHorse 1.0, per runwayml's API announcement
• Finishing and enhancement: Magnific Precision Upscaler V2, per the same post
• Access point: the developer portal, linked in runwayml's follow-up post

The most concrete signal from the evidence pool is that creators are already treating this as a pipeline, not a one-click generator. juliewdesign_'s post credits Midjourney, Nano Banana Pro, and Seedance in the same piece.

AllaAisling's post shows a similar pattern with Seedance 2.0 used inside Runway, then a 4K upscale step handled separately in Topaz. The workflow split is pretty clear:

1. Generate the base visual.
2. Push it through motion or variation.
3. Finish with an upscale pass.

That is a more interesting creative story than the raw model list, because it shows Runway's API slotting into mixed asset production instead of asking creators to stay loyal to one model family.

Two days before the API catalog post, Runway launched MCP support. runwayml's MCP launch post says the connector drops Runway into Claude, ChatGPT, Cursor, Replit, and more, with access to models including Gen-4.5, Seedance 2.0, GPT Images 2.0, and Kling.

That adds a second distribution layer on top of the API itself. Instead of opening a separate creative app, the model stack can now sit inside agent and IDE workflows, with the Runway MCP page positioned as the setup point.

The timing matters because it turns this week's API expansion into part of a larger rollout: first Runway made its generation catalog callable from agent clients, then the API announcement widened the catalog inside its own developer surface.