Qwen3.8 125B (6B active) Flash-Next

on M2 Ultra 192GB · AX Engine 7.6.0 (MLX)

Oct 10, 2026

Throughput

22.2 t/s gen · 325.3 tokens/s prompt processing (prefill)Prompt processing measures input prefill speed; it does not tell you the time to first token.

Quant

MXFP4 / affine 8-bit mixed (MLX)

MTP (Multi-Token Prediction)

System RAM

192 GB

Use cases

coding

Summary

User reports Qwen3.8 Flash-Next at 22.2 t/s generation and 325.3 t/s prefill on an M2 Ultra 192GB. Setup is AX Engine at MXFP4 / affine 8-bit mixed, speculative decoding on, one request. Medians of 20 trials in defai-digital's AutomatosX benchmark of 2026-10-08. A source-reported measurement, not independently reproduced.