oMLX 0.7.3.dev1 本地跑 M3.1 Flash：M3 Ultra 上 42 tok/s
作者实测：oQ4e 量化加 DSpark 草稿模型（Lightning MTP）后单请求从 22.5 提升到 42.0 tok/s（+87%），4K 预填充 370 tok/s、128K 仍有 298 tok/s，并给出 196/213/249GB 三档量化下载。
MiniMax M3.1 Flash runs on your Mac with oMLX 0.7.3.dev1! https://github.com/jundot/omlx/releases/tag/v0.7.3.dev1 With its bundled DSpark draft running as Lightning MTP, one request on an M3 Ultra (oQ4e) goes from 22.5 to 42.0 tok/s (+87%), New fused decode kernels then add another 6-7% with Lightning MTP! Prefill runs at 370 tok/s at 4K and still 298 tok/s at 128K. oQ builds on mlx-community - oQ3e (196GB): https://huggingface.co/mlx-community/MiniMax-M3.1-Flash-oQ3e-mtp - oQ3.5e (213GB): https://huggingface.co/mlx-community/MiniMax-M3.1-Flash-oQ3.5e-mtp - oQ4e (249GB): https://huggingface.co/mlx-community/MiniMax-M3.1-Flash-oQ4e-mtp Also in 0.7.3.dev1 - Faster M1 and M5 Macs. Qwen3.6-35B-A3B prefill on M1 Ultra goes 429 -> 644 tok/s (+50%) at 100K, and DeepSeek V4 Flash prefill on M5 Max 650 -> 767 tok/s (+18%) at 32K. - Faster DFlash on fp16 models. Qwen3.8-27B fp16 + DFlash2 on M3 Ultra goes 106 -> 141 tok/s (+33%). - Faster, leaner GLM-5.3 Flash. Two requests at 64K decode 16.4 -> 19.2 tok/s each (+17%), and the model uses 3.27 GB less memory. - Better low-bit oQ. More of the bit budget now goes to the layers every token uses, not the output head. Low-bit quants come out closer to the original model. - More models. Flint, Raon-1 mini, and MiniCPM-V 4.7 with its DSpark draft.
这条是引用着另一条写的：「MiniMax M3.1 Flash 开源权重：428B 总参、23B 激活」