Agnes-3.0-Flash
11 records across 3 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RTX PRO 6000 Blackwell Max-Q | Agnes-3.0-Flash | BF16 | SGLang | 125.6 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX PRO 6000 Blackwell Max-Q | Agnes-3.0-Flash | BF16 | SGLang | 123.7 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX PRO 6000 Blackwell Max-Q | Agnes-3.0-Flash | BF16 | SGLang | 71.6 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX PRO 6000 Blackwell Max-Q | Agnes-3.0-Flash | BF16 | SGLang | 59.0 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX PRO 6000 Blackwell Max-Q | Agnes-3.0-Flash | BF16 | SGLang | 35.2 | – | – | – | – | 2026-10-03 | community | – | source |
| M5 Max | Agnes-3.0-Flash | 8bit | oMLX | 15.8 | 631.9 | – | – | 4096 | 2026-10-18 | community | – | source |
| M1 Max | Agnes-3.0-Flash | oQ4e | oMLX | 14.9 | 61.7 | – | – | 8192 | 2026-10-09 | community | – | source |
| M5 Max | Agnes-3.0-Flash | 8bit | oMLX | 14.9 | 356.6 | – | – | 1024 | 2026-10-18 | community | – | source |
| M1 Max | Agnes-3.0-Flash | oQ4e | oMLX | 14.5 | 67.9 | – | – | 1024 | 2026-10-09 | community | – | source |
| M5 Max | Agnes-3.0-Flash | 8bit | oMLX | 12.9 | 456.9 | – | – | 4096 | 2026-10-18 | community | contradiction | source |
| M5 Max | Agnes-3.0-Flash | 8bit | oMLX | 12.6 | 329.0 | – | – | 1024 | 2026-10-18 | community | contradiction | source |