Swift 1.5 Qwen3.8-Flash-Next
34 records across 2 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 132.2 | 2194.0 | – | – | 1024 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 130.5 | 2199.0 | – | – | 1024 | 2026-10-15 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 122.4 | 4107.0 | – | – | 65536 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 121.7 | 4111.0 | – | – | 65536 | 2026-10-15 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 118.7 | 2245.0 | – | – | 1024 | 2026-10-17 | community | contradiction | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 118.5 | 4180.0 | – | – | 32768 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | 4.7bpw (4-bit experts, 8-bit spine) | MLX | 113.7 | – | – | – | 179200 | 2026-10-05 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 111.2 | 4215.0 | – | – | 16384 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 107.8 | 4010.0 | – | – | 131072 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 105.9 | 4226.0 | – | – | 16384 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 101.2 | 4010.0 | – | – | 131072 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 97.4 | 4113.0 | – | – | 65536 | 2026-10-17 | community | contradiction | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 90.8 | 3900.0 | – | – | 199680 | 2026-10-17 | community | – | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 89.9 | 884.9 | – | – | 1024 | 2026-10-12 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 88.5 | 3895.0 | – | – | 199680 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 85.8 | 3911.0 | – | – | 199680 | 2026-10-17 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 84.9 | 4022.0 | – | – | 131072 | 2026-10-17 | community | contradiction | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | 4.7bpw (4-bit experts, 8-bit spine) | MLX | 81.1 | – | – | – | 179200 | 2026-10-05 | community | – | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 79.9 | 809.7 | – | – | 4096 | 2026-10-12 | community | – | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 78.8 | 1045.0 | – | – | 4096 | 2026-10-15 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 69.2 | 4054.0 | – | – | 65536 | 2026-10-17 | community | contradiction | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 66.2 | 914.5 | – | – | 1024 | 2026-10-12 | community | contradiction | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | IQ3_XXS | llama.cpp (Metal) | 62.7 | – | – | – | 179200 | 2026-10-05 | community | – | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 60.4 | 841.4 | – | – | 4096 | 2026-10-12 | community | contradiction | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 50.8 | 993.0 | – | – | 131072 | 2026-10-10 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | IQ3_XXS | llama.cpp (Metal) | 44.7 | – | – | – | 179200 | 2026-10-05 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 32.6 | 813.4 | – | – | 1024 | 2026-10-17 | community | contradiction outlier | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 30.6 | 982.0 | – | – | 131072 | 2026-10-10 | community | contradiction | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 23.6 | 2697.0 | – | – | 16384 | 2026-10-17 | community | contradiction outlier | source |
| M3 Ultra | Swift 1.5 Qwen3.8-Flash-Next | oQ6e | oMLX | 0.8 | 1008.0 | – | – | 65536 | 2026-10-15 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | 4.7bpw (4-bit experts, 8-bit spine) | MLX | – | 3191 | – | – | 179200 | 2026-10-05 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | 4.7bpw (4-bit experts, 8-bit spine) | MLX | – | 2928 | – | – | 179200 | 2026-10-05 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | IQ3_XXS | llama.cpp (Metal) | – | 1427 | – | – | 179200 | 2026-10-05 | community | – | source |
| M5 Ultra | Swift 1.5 Qwen3.8-Flash-Next | IQ3_XXS | llama.cpp (Metal) | – | 1307 | – | – | 179200 | 2026-10-05 | community | – | source |