Vibeleaderboard
← All Intel
Intel / article

Seedance 2.5 Flash now available on AI Gateway

Source
vercel.com
Author
Zachary Chen
Date

Context

Seedance 2.5 Flash is a cheaper, faster tier of ByteDance's Seedance 2.5, a model that generates video with synchronized sound from a text prompt or from reference images, video and audio. Vercel has added it to AI Gateway, its service for calling many AI models through one API (Vercel changelog). Vercel says Flash comes close to the standard model in realistic people, reference-based editing and following the prompt, but gives no speed figure. Flash makes clips of 4 to 30 seconds, can edit or extend existing videos, and can set the first and last frames from images.

What Flash gives up shows on the two models' AI Gateway pages. Flash outputs 480p or 720p at 24 frames per second, and its playground accepts up to 10 reference images and 5 videos (Flash model page). The standard model's playground adds 1080p and accepts up to 30 images and 10 videos, and its page describes up to 50 reference inputs in one request (Seedance 2.5 model page). At the time of writing, Flash is listed at $7 per million generated , against a starting price of $10.70 for the standard model through ByteDance, about a third less per token. For work delivered at 720p or below with a handful of references, Flash is the cheaper per-token option, with output Vercel describes as close to the standard model's. Higher resolution or heavily referenced shots still need the standard model, which Vercel added in August.

Terms in this piece · Glossary
  • token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
More from Zachary Chen
Recommended reads
Comments

Checking sign-in…

Loading comments…