Alibaba to release Qwen3.8-Flash-Next open-weight MoE model

Qwen3.8-Flash-Next (~125B-A6B + 51B n-gram) is an open-weight multimodal MoE model built on the architecture for the upcoming Qwen4 family. Ideal 4-bit quant ≈ 82 GB, with real-world quants likely 80–90 GB. Unsloth announced day-0 support.