4x RTX 3090
4 records across 1 models (quants: 3.0-bit).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 4x RTX 3090 | GLM-5.3-Flash | 3.0-bit | llama.cpp (CUDA) | 32.6 | – | – | – | – | 2026-09-25 | community | – | source |
| 4x RTX 3090 | GLM-5.3-Flash | 3.0-bit | llama.cpp (CUDA) | 31.5 | – | – | – | – | 2026-09-25 | community | – | source |
| 4x RTX 3090 | GLM-5.3-Flash | 3.0-bit | llama.cpp (CUDA) | 24.7 | – | – | – | – | 2026-09-25 | community | – | source |
| 4x RTX 3090 | GLM-5.3-Flash | 3.0-bit | llama.cpp (CUDA) | 20.1 | – | – | – | – | 2026-09-25 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085