Erik Paperniuk, notes https://paperniuk.github.io/blog/ 2026-10-04T00:00:00Z Erik Paperniuk Qwen3.8-Flash-Next on a 2021 M1 Max: 44 tok/s, and 35 tok/s with 400K tokens in the context https://paperniuk.github.io/blog/flash-next-on-m1-max/ 2026-10-04T00:00:00Z A 126B mixture-of-experts model on a 64 GB laptop, with kernels written for the M1, and how little the speed falls as the context grows. Splash on M1, part 2: 35B-A3B at 144 tok/s and a head-to-head with oMLX and MTPLX https://paperniuk.github.io/blog/splash-on-m1-part-2/ 2026-09-26T00:00:00Z Attention and the MoE expert layers move off MPP too. Speed, GPU temperature, energy per token and memory against two MLX engines on the same Mac. Splash on a 2021 M1 Max: Qwen3.8-27B at 39 tok/s with custom Metal kernels https://paperniuk.github.io/blog/splash-on-m1/ 2026-09-24T00:00:00Z Splash only runs on M3 and newer. I ported it to my M1 Max and rewrote the kernels that depended on bfloat16 and Apple's MPP library, which doubled decode speed.