Qwen3-VL-30B-A3B-Instruct
10 records across 3 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M5 Max | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 114.9 | 4873.0 | – | – | 4096 | 2026-10-06 | community | – | source |
| M3 Ultra | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 104.7 | 2416.0 | – | – | 1024 | 2026-10-08 | community | – | source |
| M3 Ultra | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 97.5 | 2597.0 | – | – | 4096 | 2026-10-08 | community | – | source |
| M5 Max | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 91.6 | 4023.0 | – | – | 16384 | 2026-10-06 | community | – | source |
| M5 Pro | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 88.4 | 3132.0 | – | – | 4096 | 2026-10-16 | community | – | source |
| M5 Max | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 68.1 | 2422.0 | – | – | 32768 | 2026-10-06 | community | – | source |
| M5 Pro | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 61.2 | 2078.0 | – | – | 16384 | 2026-10-16 | community | – | source |
| M5 Max | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 43.8 | 1369.0 | – | – | 65536 | 2026-10-06 | community | – | source |
| M5 Max | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 27.4 | 776.3 | – | – | 131072 | 2026-10-06 | community | – | source |
| M5 Max | Qwen3-VL-30B-A3B-Instruct | 4bit | oMLX | 19.4 | 522.7 | – | – | 199680 | 2026-10-06 | community | – | source |