Qwen3.8 125B (6B active) Flash-Next
on M5 Air · MLX · 2,000 ctx
Summary
User reports Qwen3.8-Next running faster than a dense 27B model on an adapted DSv4 streaming stack.
Setup uses a 3-bit MoE quant against the dense model's 4-bit quant, in low power mode.
No throughput figures are given.