ByteDance's Seedance 2.5 takes video gen to 30 seconds

Anyone who has tried making video with AI knows the pain of the "clip reel" workflow: you get four seconds of gorgeous footage, stitch five of them together, and realize the main character changes faces three times.

ByteDance is trying to break out of that loop. On July 31, 2026, the company officially launched Seedance 2.5, its next-generation video creation model.

30-second continuous takes and 50 reference inputs

Seedance 2.5 stretches single-pass generation from 15 seconds to a full 30 seconds, generating both audio and video together.

More importantly, it supports multi-round extensions while preserving character consistency, visual style, and pacing. That means creators can string together multi-minute sequences without constantly patching broken transitions.

Within a single 30-second shot, the model can handle complex narrative blocking. In one demo, it follows a singer from a dressing room down a hallway and straight onto a stadium stage in a single unbroken take.

Reference capability also got a massive overhaul. You can now feed it up to 30 images, 10 video clips, and 10 audio files in one prompt—up to 50 inputs in total.

It also supports clay render referencing. Creators can block out a scene using untextured 3D models to dictate camera paths and lighting, letting the AI handle physical shading and final textures.

Editing now features timestamp-level precision alongside improved green screen and camera angle controls.

Top-tier action, but with a serious price tag

Seedance 2.5 is already live on platforms like Jimeng AI and Doubao Pro, with API access planned soon through BytePlus ModelArk.

Early reviews from creators praise it as arguably the top video generator right now, particularly highlighting its physics handling in high-action fight sequences.

That said, there are obvious trade-offs right now.

Reviewers point out the model is currently capped at 720p, though 1080p and 4K updates are reportedly in the pipeline. It is also not yet available in the US market.

It is also definitely not cheap. According to one video creator, a 10-second clip costs around 460 credits—roughly $4.60—making it noticeably pricier than rivals like MiniMax.

Why it matters

For commercial creators, indie animators, and ad agencies, Seedance 2.5 shifts AI video from an aesthetic novelty into a functional pre-visualization and production engine.

Feeding 50 reference inputs alongside 3D clay blocking solves the core friction in generative video: lack of direct directorial control. At roughly $4.60 for 10 seconds, it is steep for casual tinkering, but practically free compared to hiring a live film crew.

We think ByteDance made a smart call here. Instead of chasing hyper-realistic static clips, it prioritized shot continuity, reference tracking, and precise editing tools.

Resolution remains an issue until 1080p arrives, and the pricing will keep casual hobbyists away. But for high-end creators willing to pay, it raises the bar for what a single prompt can direct.