StepFun Launches Step 5 Preview: A 600B-Total, 27B-Active MoE Model With 1M Context for Long-Horizon Agentic Work
StepFun released Step 5 Preview, a sparse MoE model with 600B total parameters and 27B active per token. It supports a 1M-token context window and accepts text, image, and video input for long-horizon agentic tasks.
Key Takeaways
- Key Highlight：StepFun released Step 5 Preview, a sparse MoE model with 600B total parameters and 27B active per token. It supports a 1M-token context window and accepts text, image, and video input for long-horizon agentic tasks.
- Innovation & Tech：Highlights advancements in Agent, StepFun, Launches, demonstrating rapid progress in model capabilities.
- Industry Impact：Reported via MarkTechPost, offering actionable signals for developers and technology leaders.
StepFun has introduced Step 5 Preview, a sparse Mixture-of-Experts architecture designed to handle complex, long-horizon agentic workflows. With 600 billion total parameters and 27 billion active per token, the model aims to balance computational efficiency with the capacity needed for demanding applications.
A standout feature is the model's 1M-token context window, which allows it to process extensive sequences of information in a single pass. This capability is particularly relevant for software engineering and professional tasks that require maintaining context over long interactions, such as debugging large codebases or analyzing detailed documents.
Step 5 Preview also supports multimodal inputs, accepting text, image, and video data. This broadens its potential utility across various AI applications, enabling more comprehensive understanding and interaction with different types of content.
The focus on agentic work signals StepFun's intent to compete in the growing market for models that can autonomously perform multi-step tasks. By combining a large context window with efficient MoE design, the model targets use cases where sustained reasoning and action over extended periods are essential.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.
Industry Insights & Analysis
As artificial intelligence rapidly evolves, breakthroughs surrounding Agent, StepFun, Launches, Step are shifting toward scalable, robust real-world implementations.
Driven by both open-source ecosystems and proprietary model architectures, the integration between compute optimization, data engineering, and agentic workflows is accelerating. This development provides a strategic benchmark for upcoming AI tooling and developer workflows.