The product

Everything in VideoGenAI.

Not a demo, not a research preview, a working pipeline your team can rely on. Here's what's actually inside.

Inside the box

Seedance 2.0, wrapped in a workflow that ships.

✦Flagship

Direct API access to ByteDance's flagship. Multimodal inputs: text, image, audio, video. Native 1080p. 15-second clips. #1 on leaderboards.

Prompt78%

> a cinematic drone shot over snow-capped peaks

rendering…28s • 1080p

Up to 9 reference images, 3 video clips, 3 audio clips in a single prompt. Composition, motion and audio planned in one pass.

→

Seedance renders audio together with video in one pass. Your clip ships with foley, ambience and dialogue timed to the motion.

REF

Alibaba's open-source video model lands on the platform the moment the API is public. Same workspace, same tokens, one extra model in the router.

Use everything you generate in paid ads, YouTube, TikTok, client work. The license sits in our Terms, not marketing copy.

Paid adsYouTubeTikTokClients

Share brand kits, recycle tokens, collaborate on storyboards in real-time. Per-seat audit logs on Studio tier.

01
Write a prompt
Plain English. A sentence or a shot list. Attach a reference image or clip if you have one.
02
We route it
Every prompt is matched to the best-fit model in our render pool. You don't pick, you just get the right output.
03
Render in ~22s
1080p, up to 16 seconds per clip. Batch as many as you want; we queue them in parallel.
04
Export or iterate
MP4, ProRes or WebM. Keep the seed, tweak the prompt, re-render for free.