Step 3.7 Flash
StepFun's high-efficiency multimodal MoE model. 196B total parameters with 11B active, supporting text, image, and video inputs. Open-source weights available on HuggingFace.
Specifications
| Lab | StepFun |
|---|---|
| Context window | 262.1K tokens |
| Input price | $0.20/1M |
| Output price | $1.15/1M |
| Release | May 2026 |
Strengths
- Very low pricing ($0.20/M input, $1.15/M output)
- Multimodal: native image and video understanding
- 256K context window
- Open-source with HuggingFace weights
- MoE architecture for efficient inference
Weaknesses
- Only 11B active parameters limits reasoning depth
- Less proven than established frontier models
- Chinese lab potential data sovereignty concerns
Best for
- Budget multimodal tasks (image/video analysis)
- High-throughput applications where cost matters
- Developers needing open-source multimodal models
In Depth: Step 3.7 Flash
Summary
Step 3.7 Flash is an AI model from StepFun.
Released in May 2026. It currently appears in the Overall category on LMRank. It takes text input and produces text output, with a context window of 262.1K tokens. Input pricing is $0.20 per 1M tokens and output is $1.15 per 1M tokens on OpenRouter.