Can I run this LLM?
Can a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB) run Step 3.7 Flash (MoE)?
Step · Apache 2.0 · estimated for a 144.0 GB machine at 4K context.
Yes — it runs great.
Runs greatBest fit: Step 3.7 Flash (MoE) at Q8_0 · 212.0 GB weights · 48–71 tok/s on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB).
Arena Text Overall · style-controlled
UnratedNo exact matchNo reviewed exact Arena match
No reviewed exact model/configuration match in the Arena Text Overall style-controlled snapshot. No family or editorial score is substituted. Local quantization quality is unmeasured.
Snapshot published 2026-10-08 · snapshot imported 2026-10-11 UTC · validated last-good data · manual refresh
Arena leaderboard dataset · Arena / lmarena-ai, Arena Leaderboard Dataset (CC BY 4.0). Selected Text Overall style-controlled rows; field names adapted, ratings unchanged. · dataset CC BY 4.0 · revision e16f90232986
- Total memory at 4K context
- 213.1 GB
- Weights + KV cache + runtime overhead
- Estimated generation
- 48–71 tok/s
- Estimated first token
- ~0.5s
Every quantization on this machine
On small screens, each quant lists every metric below its name. On wider screens, scroll sideways for total memory, speed and fit. Keyboard: focus the table, then use the arrow keys. Quant names stay visible.
| Quant | Weights | Total memory | Speed | Fit |
|---|---|---|---|---|
| Q2_K | 71.2 GB | 72.3 GB | 257–386 tok/s | Runs great |
| Q3_K_M | 91.8 GB | 92.9 GB | 238–357 tok/s | Runs great |
| Q4_K_M | 121.6 GB | 122.7 GB | 215–322 tok/s | Runs great |
| Q5_K_M | 142.0 GB | 143.1 GB | 135–203 tok/s | Runs great |
| Q6_K | 171.8 GB | 172.8 GB | 76–114 tok/s | Runs great |
| Q8_0 | 212.0 GB | 213.1 GB | 48–71 tok/s | Runs great |
Ran Step 3.7 Flash (MoE) on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB)? Add your real numbers →