Tag
The author released mainline-compatible imatrix quants for the Qwen3.8-Flash-Next model, offering 20–30GB smaller sizes with competitive PPL, and includes a separate ROCm build for AMD hardware.