🔌
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 Intel Gaudi 3 run LLaVA-CoT?
Yes — here's how fast and up to what context, with real measured numbers.
LLaVA-CoT on the Intel Gaudi 3
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 575k | 97.5-121.1 t/s | ~1.1-1.3 (slow) |
| Q8 | yes | 642k | 195.1-242.2 t/s | ~2.1-2.6 (slow) |
| Q4 | yes | 672k | 348.4-432.5 t/s | ~3.8-4.7 (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 Intel Gaudi 3 run LLaVA-CoT?▶
Yes. At Q4 it generates ~348.4-432.5 tok/s and fits up to 672k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~3.8-4.7 tok/s.
How much context fits?▶
At Q4, up to about 672k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.