can-i-run-this-llm

Can I run this LLM?

Can a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB) run GLM-4.7 Flash (MoE)?

GLM · MIT · estimated for a 144.0 GB machine at 4K context.

Yes — it runs great.

Best fit: GLM-4.7 Flash (MoE) at FP16 · 59.9 GB · 222–333 tok/s on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K11.3 GB317–475 tok/sRuns great
Q3_K_M14.6 GB308–462 tok/sRuns great
Q4_K_M18.3 GB299–448 tok/sRuns great
Q5_K_M21.4 GB291–437 tok/sRuns great
Q6_K24.7 GB284–426 tok/sRuns great
Q8_031.8 GB269–403 tok/sRuns great
FP1659.9 GB222–333 tok/sRuns great

Ran GLM-4.7 Flash (MoE) on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB)? Add your real numbers →