- Although pricing has been announced, the Volcengine website indicates that the model does not yet support widespread API access.
- The model is currently restricted to internal use and an initial group of invited enterprise clients.
ByteDance's cloud computing arm, Volcengine, has unveiled the API pricing for its new-generation multimodal video generation model, Seedance 2.0.
The cost for pure video generation is 46 yuan ($6.66) per million tokens, while video editing is priced at 28 yuan, according to the Volcengine website.
Generating a 15-second video consumes about 308,880 tokens. This translates to a pure video generation cost of about 1 yuan per second.
Despite the published pricing, the Volcengine website shows that the model is temporarily unavailable for broad API access.
Some Chinese media reports suggest that Seedance 2.0 is currently closed to third-party tool developers and is restricted to ByteDance's internal use.
Furthermore, the model has only been integrated by an initial batch of enterprise clients, including a leading comic-to-animation company.
Officially launched on February 12 this year, Seedance 2.0 adopts a unified multimodal architecture for joint audio and video generation.
It supports four input modalities — text, image, audio, and video — and is capable of generating 15-second, high-quality, multi-shot audio-visual content.
However, the powerful AI tool is facing severe copyright challenges. In February, The Walt Disney Company accused ByteDance of using its works to train the model without permission.
Subsequently, ByteDance's Japanese subsidiary announced adjustments to the Seedance 2.0 service. The move aims to prevent users from generating videos that infringe on the intellectual property rights of Disney, "Ultraman," and other related IPs.
($1 = 6.8974 yuan)
