🔌
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 Qwen-VL?
Yes — here's how fast and up to what context, with real measured numbers.
Qwen-VL on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 832k | 182.5-217.5 t/s | ~1.5-1.8 (slow) |
| Q8 | yes | 905k | 365-435 t/s | ~3-3.6 (slow) |
| Q4 | yes | 937k | 651.8-776.8 t/s | ~5.4-6.5 (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 Qwen-VL?▶
Yes. At Q4 it generates ~651.8-776.8 tok/s and fits up to 937k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~5.4-6.5 tok/s.
How much context fits?▶
At Q4, up to about 937k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.