Kinovi AI

Seedance 2.1 is Kinovi’s upcoming route for ByteDance’s next-generation video model, designed to move from simple generation toward more production-ready creation.

It’s built for richer multimodal inputs—up to 20 images, up to 6 audio files, and up to 6 video clips in a single generation—so you can bring more creative context (characters, product angles, style references, and more) into one request. The model is also expected to support up to 30-second outputs and maintain smoother motion and stronger scene consistency across longer runs.

For more precise revisions, Seedance 2.1 introduces frame-level correction and improved prompt adherence, helping complex instructions translate more reliably into the final video. It also targets enhanced character consistency and improved multilingual support for global creative workflows.

What can I generate with Seedance 2.1 on Kinovi?

You can create multimodal AI videos using a mix of text plus reference inputs such as images, audio files, and video clips, with support for longer clips up to 30 seconds.

What is the maximum video duration for Seedance 2.1?

Seedance 2.1 is expected to support videos from 4 to 30 seconds.

How many reference assets can I include in one Seedance 2.1 generation?

A single generation can include up to 20 reference images, up to 6 reference video clips, and up to 6 audio files.

What does “frame-level editing” mean for Seedance 2.1 workflows?

It’s intended to make it easier to correct specific moments instead of regenerating the entire clip when small issues appear (for example facial expression, hand motion, or camera transitions).

Which generation modes does Seedance 2.1 support?

Seedance 2.1 includes a reference mode for multimodal generation, and a keyframe mode that generates using the first two imageUrls as the start and end frames.

When will Seedance 2.1 be available?

Availability is expected in late June to early July 2026; until then, Seedance 2.0 remains the current production baseline.