ByteDance Launches Seedance 2.5 Video Model API with Native 30-Second Generation

ByteDance's Volcano Engine has officially launched the Seedance 2.5 video model API, which natively supports generating 30-second videos. Compared to version 2.0, it features significant improvements in instruction following, long narrative capabilities, realism, and support for up to 50 multi-modal reference assets. The ability to natively generate 30-second videos marks a major advancement in AI video generation, enabling creators to produce longer, more coherent narrative content. By offering this via API, ByteDance makes high-quality, long-form AI video generation more accessible to developers and enterprise platforms. The API pricing is set at 42 RMB per million tokens for inputs containing video and 70 RMB per million tokens for inputs without video. Additionally, the model supports over ten languages and is already integrated into platforms like LibTV and TapNow.