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 TiTok-L?
Yes — here's how fast and up to what context, with real measured numbers.
TiTok-L on the NVIDIA B200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 2663k | 9511.4-11335.5 t/s | ~47.6-56.7 (slow) |
| Q8 | yes | 2668k | 19022.8-22671 t/s | ~95.1-113.4 (slow) |
| Q4 | yes | 2670k | 33969.3-40483.9 t/s | ~169.8-202.4 (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 TiTok-L?▶
Yes. At Q4 it generates ~33969.3-40483.9 tok/s and fits up to 2670k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~169.8-202.4 tok/s.
How much context fits?▶
At Q4, up to about 2670k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.