Loading...
Loading...
ByteDance Seed is the foundation-model research team of ByteDance, founded in 2023, reporting to ByteDance CEO Liang Rubo and spread across offices in Beijing, Shanghai, Shenzhen and Hangzhou. It develops the Seed language models (marketed as Doubao Seed) behind the Doubao assistant, the Seedream image models, the Seedance video models and the OmniHuman human-animation models. The Doubao models power more than 50 application scenarios across ByteDance products.
first Seedream text-to-image model
second generation
deeper instruction understanding, real-time search
ByteDance's flagship Seedream with deep-thinking prompt understanding and native text in 14 languages, up to 2K.
A fast Seedream tier for workflows where speed and budget matter, up to 2K.
cost-optimised 1.0 tier
multi-shot 1080p text and image to video
joint audio-visual generation
text, image, audio and video input; multi-shot storytelling
faster 2.0 endpoint
30-second clips from up to 50 multimodal references