Qwen3-4B-Instruct-2507
43 records across 8 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M5 Max | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 141.5 | 4948.0 | – | – | 1024 | 2026-10-15 | community | – | source |
| M4 Max | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 136.2 | 1554.0 | – | – | 1024 | 2026-10-14 | community | – | source |
| M5 Max | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 123.8 | 4811.0 | – | – | 4096 | 2026-10-15 | community | – | source |
| M4 Max | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 119.8 | 1396.0 | – | – | 4096 | 2026-10-14 | community | – | source |
| M1 Max | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 94.8 | 734.1 | – | – | 1024 | 2026-10-13 | community | – | source |
| M5 Max | Qwen3-4B-Instruct-2507 | 8bit | oMLX | 83.2 | 4456.0 | – | – | 1024 | 2026-10-09 | community | – | source |
| M1 Max | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 81.2 | 670.5 | – | – | 4096 | 2026-10-13 | community | – | source |
| M5 Max | Qwen3-4B-Instruct-2507 | 8bit | oMLX | 78.4 | 4451.0 | – | – | 4096 | 2026-10-09 | community | – | source |
| M1 Max | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 76.5 | 551.2 | – | – | 1024 | 2026-10-07 | community | – | source |
| 4070 laptop card | Qwen3-4B-Instruct-2507 | Q4_K_M | llama.cpp | 72 | – | – | – | – | 2026-09 | community | – | source |
| M5 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 48.2 | 1651.0 | – | – | 1024 | 2026-10-15 | community | – | source |
| M5 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 47.9 | 1497.0 | – | – | 1024 | 2026-10-15 | community | – | source |
| M5 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 47.9 | 1497.0 | – | – | 1024 | 2026-10-15 | community | – | source |
| M5 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 39.9 | 1480.0 | – | – | 4096 | 2026-10-15 | community | – | source |
| M5 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 39.9 | 1480.0 | – | – | 4096 | 2026-10-15 | community | – | source |
| M5 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 39.6 | 1543.0 | – | – | 4096 | 2026-10-15 | community | – | source |
| M4 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 37.9 | 420.8 | – | – | 1024 | 2026-10-08 | community | – | source |
| M4 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 31.8 | 381.9 | – | – | 4096 | 2026-10-08 | community | – | source |
| M4 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 31.7 | 390.7 | – | – | 4096 | 2026-10-12 | community | – | source |
| M1 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 23.0 | 184.1 | – | – | 1024 | 2026-10-16 | community | – | source |
| M1 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 22.9 | 184.2 | – | – | 1024 | 2026-10-16 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | 20.3 | – | – | – | 128 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | 20.22 | – | – | – | 128 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | 20.05 | – | – | – | 128 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | 19.97 | – | – | – | 128 | 2026-10 | community | – | source |
| M1 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 19.5 | 167.7 | – | – | 4096 | 2026-10-16 | community | – | source |
| M1 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 19.4 | 167.3 | – | – | 4096 | 2026-10-16 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | 19.21 | – | – | – | 128 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | 19.14 | – | – | – | 128 | 2026-10 | community | – | source |
| M4 | Qwen3-4B-Instruct-2507 | 4bit | oMLX | 16.9 | 274.5 | – | – | 16384 | 2026-10-12 | community | – | source |
| 4070 laptop card | Qwen3-4B-Instruct-2507 | Q4_K_M | llama.cpp | 11.4 | – | – | – | – | 2026-09 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 156.0 | – | – | 2048 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 262.06 | – | – | 2048 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 281.31 | – | – | 2048 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 474.37 | – | – | 2048 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 451.55 | – | – | 2048 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 741.54 | – | – | 2048 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 135.97 | – | – | 4096 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 208.94 | – | – | 4096 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 246.41 | – | – | 4096 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 381.71 | – | – | 4096 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 401.77 | – | – | 4096 | 2026-10 | community | – | source |
| AMD Ryzen 9 9950X | Qwen3-4B-Instruct-2507 | Q4_0 | llama.cpp (CPU) | – | 610.93 | – | – | 4096 | 2026-10 | community | – | source |