can-i-run-this-llm

Can I run this LLM?

Can a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB) run Ling 3.0 Flash (MoE)?

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

Yes — it runs great.

Best fit: Ling 3.0 Flash (MoE) at FP16 · 255.1 GB · 46–68 tok/s on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K45.9 GB308–462 tok/sRuns great
Q3_K_M59.0 GB295–443 tok/sRuns great
Q4_K_M77.8 GB278–417 tok/sRuns great
Q5_K_M90.7 GB267–401 tok/sRuns great
Q6_K109.3 GB253–380 tok/sRuns great
Q8_0135.6 GB195–292 tok/sRuns great
FP16255.1 GB46–68 tok/sRuns great

Ran Ling 3.0 Flash (MoE) on a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB)? Add your real numbers →