🔌
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 Unicorn?
Yes — here's how fast and up to what context, with real measured numbers.
Unicorn on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 648k | 159.3-189.8 t/s | ~1.3-1.6 (slow) |
| Q8 | yes | 715k | 318.5-379.6 t/s | ~2.7-3.2 (slow) |
| Q4 | yes | 745k | 568.8-677.9 t/s | ~4.7-5.6 (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 Unicorn?▶
Yes. At Q4 it generates ~568.8-677.9 tok/s and fits up to 745k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~4.7-5.6 tok/s.
How much context fits?▶
At Q4, up to about 745k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.