Step 3.7 Flash
Step 3.7 Flash is an open-weights language model from StepFun, released May 29, 2026. It has 201B parameters (dense). The context window is 250K tokens. API price: $0.2 input / $1.21 output per million tokens.
Step 3.7 Flash is a 198B-parameter sparse Mixture-of-Experts (MoE) vision-language model that combines a 196B-parameter language backbone with a 1.8B-parameter vision encoder for native image understanding.
Check it against your own PC →
Can you run Step 3.7 Flash locally?
Even with 32 GB of VRAM and 32 GB of RAM it is too large to run at a usable speed on a single consumer GPU. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Too large | – | – | – | – |
| 8 GB VRAM | Too large | – | – | – | – |
| 12 GB VRAM | Too large | – | – | – | – |
| 16 GB VRAM | Too large | – | – | – | – |
| 24 GB VRAM | Too large | – | – | – | – |
| 32 GB VRAM | Very slow | IQ1_M (experimental) | 31.2 GB | 20.5 GB | ~1 |
Ready-made quantizations on Hugging Face: unsloth/Step-3.7-Flash-GGUF · stepfun-ai/Step-3.7-Flash-GGUF · bartowski/Step-3.7-Flash-GGUF · mudler/Step-3.7-Flash-APEX-GGUF · vultr/Step-3.7-Flash-GGUF · EspenJD/Step-3.7-Flash-Q4_K_M-MTP-GGUF
Benchmark results
Epoch AI has not published results for Step 3.7 Flash yet. New models are usually evaluated within days to weeks.
Links
Announcement · Hugging Face: stepfun-ai/Step-3.7-Flash