🔌
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 TinyLlama-1.1B (1T token checkpoint)?
Yes — here's how fast and up to what context, with real measured numbers.
TinyLlama-1.1B (1T token checkpoint) on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1923k | 1592.7-1898.2 t/s | ~13.3-15.8 (slow) |
| Q8 | yes | 1940k | 3185.5-3796.4 t/s | ~26.5-31.6 (slow) |
| Q4 | yes | 1947k | 5688.3-6779.2 t/s | ~47.4-56.5 (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 TinyLlama-1.1B (1T token checkpoint)?▶
Yes. At Q4 it generates ~5688.3-6779.2 tok/s and fits up to 1947k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~47.4-56.5 tok/s.
How much context fits?▶
At Q4, up to about 1947k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.