🔌
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 SambaLingo-Thai-Chat (7B)?
Yes — here's how fast and up to what context, with real measured numbers.
SambaLingo-Thai-Chat (7B) on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 872k | 252.1-300.4 t/s | ~2.1-2.5 (slow) |
| Q8 | yes | 925k | 504.2-600.9 t/s | ~4.2-5 (slow) |
| Q4 | yes | 949k | 900.3-1073 t/s | ~7.5-8.9 (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 SambaLingo-Thai-Chat (7B)?▶
Yes. At Q4 it generates ~900.3-1073 tok/s and fits up to 949k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~7.5-8.9 tok/s.
How much context fits?▶
At Q4, up to about 949k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.