ByteDance’s Seedance 2.0 spent the spring on top of the video-model leaderboards. The variant that may matter more to working creators hasn’t even shipped yet.
It’s called Seedance 2.0 Mini, and it breaks the rule every model family has followed: the newest member is supposed to be the most expensive. Mini is reported to be the cheapest entry in the whole Seedance line, priced below even the Fast tier, while keeping quality closer to the standard model. If both halves of that claim survive launch, the economics of short-form AI video change for anyone rendering more than a few clips a week.
Where Mini sits in the family
The standard Seedance 2.0 is the multimodal generator that climbed past Kling, Veo, and Sora this year: text, images, reference video, and audio go in as one instruction; clips come out at up to 1080p, 4 to 15 seconds, with characters that stay consistent across shots. Fast is the cheaper second rung with some premium modes removed. Mini is the third: in plain terms, a lighter build of the same architecture, tuned for cost rather than ceiling.
The price ladder
- Seedance 2.0: roughly $0.096 per second at the promotional rate
- Seedance 2.0 Fast: roughly $0.076 per second
- Seedance 2.0 Mini: reported at about $0.073 per second
The gap looks tiny. The direction is the story: a brand-new model arriving below the family’s floor, while reportedly beating the tier above it on quality. One caveat — the figure traces to a single May report, echoed by Chinese AI media and a matching RMB number on Reddit. Corroboration, not confirmation.
Confirmed versus reported
Confirmed by ByteDance: nothing. No blog post, no model card. Reported with multiple signals: the pricing, the lighter-than-Fast positioning, an unchanged API surface, an early-summer window. Reported by one source only: “outperforms Fast.” The open question with real consequences is whether Mini keeps the face and dialogue modes that Fast strips; if not, character-driven work stays on the standard model regardless of price.
How to get ready
You can’t generate with Mini anywhere yet. What you can do is build on the family it belongs to, so launch day is a dropdown change rather than a migration. That’s the approach seedance2mini.ai takes: the released Seedance 2.0 and Fast models run today, text-to-video, image-to-video, first-and-last-frame, and reference-guided generation up to 1080p, on credits, with Mini already in the model selector marked “Soon.” Iterate cheap at 480p, promote the winning prompt to 1080p, and keep your eval prompts saved for the day a new tier appears.
The bottom line
Mini is well-corroborated rumor with one credible detail: it comes from the company that already took the top of the leaderboard, and the leak pattern matches how Seedance 2.0 itself surfaced before launch. Every number above except the live ones carries a “reported” asterisk. But the cheapest member of the best-scoring video family is worth being ready for, and being ready costs nothing.
Discover more from WikiTechLibrary
Subscribe to get the latest posts sent to your email.
