Step 3.7 Flash

Step 3.7 Flash is a 198B-parameter vision-language MoE model from StepFun, released in May 2026. It is accessible globally via platform.stepfun.ai, OpenRouter, or NVIDIA NIM; call the API via OpenAI-compatible endpoints with the model ID `step-3.7-flash`. The model is best suited for coding agents, agentic workflows, image processing, and search with tool use. It features a 256K-token context window and generation speeds of ~356 tokens/second. Pricing is competitive at $0.20/$1.15 per million input/output tokens. For best results, combine image and text inputs and design agentic prompts using function or tool calling.