🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA H200 run MOSS-Moon-003?
Yes — here's how fast and up to what context, with real measured numbers.
MOSS-Moon-003 on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 367k | 109.5-130.5 t/s | ~0.9-1.1 (slow) |
| Q8 | yes | 428k | 219-261 t/s | ~1.8-2.2 (slow) |
| Q4 | yes | 455k | 391.1-466.1 t/s | ~3.3-3.9 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to system RAM offload (depends on your system RAM) and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA H200 run MOSS-Moon-003?▶
Yes. At Q4 it generates ~391.1-466.1 tok/s and fits up to 455k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~3.3-3.9 tok/s.
How much context fits?▶
At Q4, up to about 455k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.