StepFun
Its own Step foundation models for language, vision, audio and videoReference profile
About StepFun
StepFun builds its own foundation models, the Step series, covering language, multimodal understanding, audio and video. It was founded in April 2023.
Models such as Step-3.5-Flash, Step-Audio 2 and Step-Video-T2V are released with open weights on GitHub and Hugging Face (stepfun-ai), and its open platform offers an API.
Why it matters
StepFun builds multimodal models across text, speech, image and video, some released as open weights.
Updates & reading
No articles or updates yet. Claim it and connect a feed.