4× Tesla P40
5 records across 2 models (quants: UD-Q3_K_XL).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 4× Tesla P40 | qwen3.5moe 122B-A10B 122B | UD-Q3_K_XL | llama.cpp (CUDA) | 29.5 | – | – | – | – | 2026-08 | community | – | source |
| 4× Tesla P40 | Qwen3.8-Flash-Next 125B | UD-Q3_K_XL | llama.cpp (CUDA) | 18.1 | – | – | – | – | 2026-08 | community | – | source |
| 4× Tesla P40 | Qwen3.8-Flash-Next 125B | UD-Q3_K_XL | llama.cpp (CUDA) | 18.1 | – | – | – | – | 2026-08 | community | – | source |
| 4× Tesla P40 | Qwen3.8-Flash-Next 125B | UD-Q3_K_XL | llama.cpp (CUDA) | 18.0 | – | – | – | – | 2026-08 | community | – | source |
| 4× Tesla P40 | Qwen3.8-Flash-Next 125B | UD-Q3_K_XL | llama.cpp (CUDA) | 18.0 | – | – | – | – | 2026-08 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085