ByteDance: Seedance 2.0 Mini

ByteDance: Seedance 2.0 Mini is a video generation model from ByteDance, the company behind the Seedance family of generative video systems. It sits in the video category on api14, and its job is to turn prompts and reference material into finished video clips. It is the smaller sibling in the Seedance 2.0 line, aimed at developers who want ByteDance video generation without reaching for the largest variant.

The model accepts text, image, video and audio as inputs, and it returns video as its only output. That combination covers several generation modes: text-to-video from a written prompt, image-to-video with control over the first frame, the last frame, or both, and multimodal reference-to-video where images, video and audio guide the result. Output resolutions are 480p and 720p, and clips can run from 4 to 15 seconds in one-second steps. Aspect ratios cover 1:1, 3:4, 9:16, 4:3, 16:9, 21:9, 9:21. Audio generation is supported, so a clip can carry a generated soundtrack rather than arriving silent. No context length is published for this model.

Those capabilities point to a fairly specific set of jobs. Short vertical clips at 9:16 for social feeds, square 1:1 cuts for in-feed placements, and wide 16:9 or 21:9 frames for web and presentation use are all directly supported by the aspect ratio list. First and last frame control suits shot-to-shot transitions, product reveals, or any case where you need a clip to start on one still and land on another. Reference-to-video with image, video and audio inputs fits style transfer, character or scene consistency across shots, and music-driven edits where the audio should shape the pacing. The 4 to 15 second range keeps outputs in the territory of individual shots and short sequences rather than long-form narrative.

You call it through api14's OpenAI-compatible API using the model id "bytedance/seedance-2.0-mini". Because the interface follows the OpenAI schema, existing clients and SDKs work with a change of model id and endpoint. Among the parameters listed for this model, frequency_penalty is supported. The generation-specific controls, resolution, aspect ratio, duration, first and last frame images, and audio generation, are part of the model's request surface rather than the shared chat parameters.

There are a few things to weigh before committing. No context length is published, and max output is listed as 0, so token-based reasoning about prompt size is not useful here; treat the prompt as a creative brief rather than a long document. Output tops out at 720p, so work that needs higher resolution, or clips longer than 15 seconds, may fit a different model better. If your task is text or image generation rather than video, this model is the wrong category entirely.