can-i-run-this-llm

Can I run this LLM?

Can a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB) run DeepSeek-R1-Distill 32B?

DeepSeek · Apache 2.0 · estimated for a 144.0 GB machine at 4K context.

Yes — it runs great.

Best fit: DeepSeek-R1-Distill 32B at FP16 · 65.5 GB · 47–71 tok/s on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K12.3 GB156–233 tok/sRuns great
Q3_K_M15.9 GB135–202 tok/sRuns great
Q4_K_M19.9 GB118–176 tok/sRuns great
Q5_K_M23.3 GB106–159 tok/sRuns great
Q6_K26.9 GB96–143 tok/sRuns great
Q8_034.8 GB79–119 tok/sRuns great
FP1665.5 GB47–71 tok/sRuns great

Ran DeepSeek-R1-Distill 32B on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB)? Add your real numbers →