StepFun is a Chinese foundation-model company behind the Step model family and the Yuewen assistant, spanning chat, multimodal understanding, and industry deployments.
StepFun builds general-purpose AI foundation models. The Step family covers language, vision, and multimodal understanding with long-context and complex reasoning support. On the consumer side, the company ships the Yuewen assistant across web and mobile for Q&A, writing, translation, and image understanding; on the enterprise side, an open platform provides model APIs and solutions for finance, automotive, and content creation. Founded by former Microsoft and ByteDance technical leaders, StepFun is one of China’s prominent independent model labs.
For developers, the Step APIs attract with stable Chinese-language performance and long-context handling suited to knowledge Q&A and document analysis. For everyday users, Yuewen consolidates search, writing, and image understanding into a single entry point that feels natural in daily use.
Developers often test Step models on long-text work such as knowledge-base Q&A and contract analysis, where the long context advantage shows. Yuewen serves everyday search and writing, with source citations boosting answer credibility. Enterprises can customize models and deployment per industry with predictable rollout timelines.
Enterprise engagements include evaluation support, helping teams benchmark model performance on their own data before committing. The consumer assistant doubles as a low-risk testing ground for capabilities that later arrive in the API.
Yuewen’s basics are free; open platform calls are billed by tokens, and private deployments are quoted per enterprise engagement.
Developers and enterprises needing high-quality Chinese LLM APIs, plus individuals wanting a capable AI assistant. It suits development teams and enterprises needing Chinese LLM APIs.