🔌
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 GLM-10B-unidirectional?
Yes — here's how fast and up to what context, with real measured numbers.
GLM-10B-unidirectional on the NVIDIA B200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 947k | 292-348 t/s | ~1.5-1.7 (slow) |
| Q8 | yes | 1008k | 584-696 t/s | ~2.9-3.5 (slow) |
| Q4 | yes | 1035k | 1042.9-1242.9 t/s | ~5.2-6.2 (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 GLM-10B-unidirectional?▶
Yes. At Q4 it generates ~1042.9-1242.9 tok/s and fits up to 1035k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~5.2-6.2 tok/s.
How much context fits?▶
At Q4, up to about 1035k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.