2 x RTX 3060
1 records across 1 models (quants: IQ4_XS).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2 x RTX 3060 | Ornith-1.5-35B-A3B | IQ4_XS | llama.cpp (CUDA) | 120 | 1800 | – | – | – | 2026-10-02 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085