New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the NVIDIA B200 run TurboVLA?
Yes — here's how fast and up to what context, with real measured numbers.
TurboVLA on the NVIDIA B200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 2666k | 14600-17400 t/s | ~73-87 (slow) |
| Q8 | yes | 2669k | 29200-34800 t/s | ~146-174 (slow) |
| Q4 | yes | 2671k | 52142.9-62142.9 t/s | ~260.7-310.7 (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 TurboVLA?▶
Yes. At Q4 it generates ~52142.9-62142.9 tok/s and fits up to 2671k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~260.7-310.7 tok/s.
How much context fits?▶
At Q4, up to about 2671k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.