Wire flash
StepFun releases Step 5 Preview AI model on OpenRouter with 1M context and multimodal input
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
StepFun has released Step 5 Preview, its new flagship AI model for agentic work, now available on the OpenRouter platform. The model uses a sparse mixture-of-experts (MoE) architecture with 27 billion active parameters out of 600 billion total, and supports a 1 million token context window. It accepts text, image, and video inputs. According to the announcement, the model is strong at coding and professional knowledge work, particularly in finance. The release positions Step 5 Preview as a competitor in the large language model space, emphasizing its multimodal capabilities and long-context processing for complex agentic tasks.
Source report
StepFun has launched the Step 5 Preview on OpenRouter, introducing a new flagship model designed for agentic workflows.
Key Specifications
- Architecture: Sparse Mixture of Experts (MoE)
- 27B active parameters
- 600B total parameters
- Context Window: 1 million tokens
- Input Modalities: Text, image, and video
Performance Highlights
The model demonstrates strong capabilities in:
- Coding
- Professional knowledge work, particularly in finance
Try It
Access the Step 5 Preview here: https://t.co/kQR1mciHSv
Source
OpenRouterNeutral / independent