Vercel Adds Seedance 2.5 Multimodal Video Model to AI Gateway.

Vercel is adding ByteDance’s Seedance 2.5 to its AI Gateway, giving developers a new option for generating, extending, and editing video inside AI-powered applications. The model can create clips up to 30 seconds long while maintaining camera movement and visual continuity, reducing the need to assemble separate shots during post-production.

The company announced the integration in its changelog on August 6, 2026. Seedance 2.5 is available through the AI Gateway, Vercel’s service for accessing AI models through a shared platform rather than building a separate provider integration for each one.

Seedance 2.5 accepts text, images, video, and audio as inputs within the same request. That allows a developer to combine characteristics from multiple source assets, such as taking a subject’s appearance from an image while borrowing motion or camera direction from an existing video.

The model’s main capabilities include:

• Video generation: It can produce a single clip of up to 30 seconds, with camera movement and continuity maintained throughout the sequence.
• Clip extension: Existing short clips can be extended while carrying forward the character, scene, and camera movement.
• Multimodal prompting: Text, video, image, and audio can be supplied together in one request.
• Local video editing: Seedance 2.5 can replace backgrounds, products, or characters while preserving the original framing, camera movement, and pacing.
• Multilingual prompts: Vercel says the model understands prompts in more than ten languages.

The local editing capability is particularly relevant for teams producing variations of the same creative. Instead of regenerating an entire video, an application could change a featured product, character, or setting while retaining the structure of the original cut. That could support localized advertising, product demonstrations, social media variants, or personalized campaign assets, although output quality will still depend on the source material and prompts.

For developers, the practical change is model access through an existing Vercel service. Applications already using AI Gateway can potentially add video generation without establishing a separate direct integration with ByteDance. Vercel’s announcement does not provide implementation code, pricing details, rate limits, or processing times, so teams should check the Seedance 2.5 model page before planning production workloads.

The release also broadens AI Gateway beyond conventional text generation. Video generation generally involves larger inputs and outputs, longer-running requests, and different user-interface requirements than a chatbot. Builders will need to design for upload handling, generation status, output storage, moderation, and failed requests rather than treating the model as a drop-in text endpoint.

Seedance 2.5 arrives as Vercel continues expanding the range of media and agent models available through AI Gateway. The addition sits alongside Vercel’s recent support for Grok Imagine Image 2.0, an image generation and editing model, and its AI Gateway integration with Hermes Agent.

For small teams, the appeal is consolidation. A product could route text, image, agent, and now more advanced video workflows through the same gateway layer. Still, the announcement is focused on model availability, not a complete video-production system. Editing interfaces, approval workflows, asset management, and publishing remain responsibilities for the application developer.

What is Vercel Seedance 2.5?

Seedance 2.5 is ByteDance’s multimodal AI video model, now available through Vercel AI Gateway. It can generate videos up to 30 seconds, extend existing clips, and edit selected elements while preserving framing, movement, and pacing.

What inputs does Seedance 2.5 support?

The model accepts text, image, video, and audio inputs in the same request. It can use different assets as references for a subject’s appearance, motion, scene, or camera work.

When is Seedance 2.5 available on Vercel?

Vercel announced availability through AI Gateway on August 6, 2026. The changelog does not specify separate pricing, plan requirements, rate limits, or regional restrictions for the model.