On November 12, Volcano Engine launched the video generation model Seedance2.5flash, positioned as the high cost-performance version whose results are currently closest to the flagship Seedance2.5, with generation speed doubled, targeting batch content production. Taking 720P text-to-video and image-to-video as an example, the generation price is about 0.99 yuan/second, roughly 34% lower than Seedance2.5.
A single segment supports video generation of up to 30 seconds, with up to 10 images, 5 video clips, and 5 audio clips per input, offering two resolution tiers: 480P and 720P. It will gradually roll out starting today on platforms such as Jimeng, LibTV, Updream, Creao AI, and TapNow, and the API will go live on Volcano Ark this Sunday (November 15).
Officially, Seedance2.5flash delivers overall results superior to the previous flagship Seedance2.0, maintaining a high standard in instruction following, physical motion, and audio performance, with natural and textured visuals; its video editing capability supports producing multiple versions around the same creative concept.
Comparing with the same telescope ad Prompt, the two finished videos are close in effect: Seedance2.5 presents the close-up of the boy adjusting the telescope more completely, while flash focuses more on completing the continuous action line of "eye to eyepiece observation—putting down the telescope—smiling in response"; the original article also cites cases such as a one-take circus video and a 30-second, 14-shot headphone ad, pointing toward batch production of film and advertising material.
Since the release of Seedance2.5, the team has successively launched the Draft sample mode, 1080P, and the upcoming 4K resolution, and now adds the flash member, forming a dual-tier structure of flagship and flash: the flagship Seedance2.5 represents the current strongest intelligence level, and combined with the Draft mode's "pick samples first, then generate the final cut" it reduces the high-definition generation costs of repeated trial and error, serving 1080P and 4K production; flash, with near-flagship results, faster speed, and lower cost, handles high-frequency creation and batch production.
Video generation competition is shifting from a contest of image quality to unit cost and output efficiency; lightweight versions taking on large-scale production while the flagship holds the quality ceiling has become a common way of organizing model matrices.