4 A10
12 records across 1 models (quants: AWQ).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 4 A10 | Qwen7B-awq | AWQ | SGLang | 2560.58 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | SGLang | 2540.23 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | SGLang | 2526.98 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | SGLang | 2123.63 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | SGLang | 2119.20 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | SGLang | 2086.65 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | vLLM | 1914.26 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | vLLM | 1887.62 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | vLLM | 1566.10 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | vLLM | 1534.41 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | vLLM | 1318.94 | – | – | – | 8192 | 2025-04 | community | – | source |
| 4 A10 | Qwen7B-awq | AWQ | vLLM | 1309.40 | – | – | – | 8192 | 2025-04 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085