🔌
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 B200 run CogVLM-17B?
Yes — here's how fast and up to what context, with real measured numbers.
CogVLM-17B on the NVIDIA B200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 538k | 171.8-204.7 t/s | ~0.9-1 (slow) |
| Q8 | yes | 603k | 343.5-409.4 t/s | ~1.7-2 (slow) |
| Q4 | yes | 632k | 613.4-731.1 t/s | ~3.1-3.7 (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 B200 run CogVLM-17B?▶
Yes. At Q4 it generates ~613.4-731.1 tok/s and fits up to 632k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~3.1-3.7 tok/s.
How much context fits?▶
At Q4, up to about 632k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.