4x B300
10 records across 1 models (quants: NVFP4).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 51.5 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 142.7 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 144.1 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 146.0 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 62.2 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 39.7 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 62.2 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 59.0 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 102.1 | – | – | – | 2026-10 | community | – | source |
| 4x B300 | GLM-5.3 | NVFP4 | vLLM | – | 102.0 | – | – | – | 2026-10 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085