can-i-run-this-llm

Rig report

NVIDIA GH200 Grace Hopper (144GB HBM + 480GB)

624.0 GB unified memory · 4900 GB/s · 18 of 18 tracked models fit

Best model it can run

Kimi K2.7 (MoE)Runs great

Q3_K_M · 442.6 GB · 26–39 tok/s · capability 92/100

Fastest that fits: Llama 3.2 1B at 290–435 tok/s

Also runs

Open the full calculator →Check another machine

Have a NVIDIA GH200 Grace Hopper (144GB HBM + 480GB)? Add your real benchmark → to sharpen these numbers for everyone.

Estimates from device specs and model architectures, refined by community benchmarks. See the methodology.