🔌
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 InternVL2_5-38B?
Yes — here's how fast and up to what context, with real measured numbers.
InternVL2_5-38B on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 196k | 45.6-54.4 t/s | ~0.4-0.5 (slow) |
| Q8 | yes | 343k | 91.3-108.8 t/s | ~0.8-0.9 (slow) |
| Q4 | yes | 407k | 162.9-194.2 t/s | ~1.4-1.6 (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 InternVL2_5-38B?▶
Yes. At Q4 it generates ~162.9-194.2 tok/s and fits up to 407k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1.4-1.6 tok/s.
How much context fits?▶
At Q4, up to about 407k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.