Support for QWEN3.8-Flash-Next
Author: doublerobustCreated Aug 26, 2026Updated Sep 16, 2026
125B params plus 51B n-gram embeddings, only 6B active per token. Looks like an ideal model for local deployment and a serious contender to deepseek-v4-flash.
https://www.modelscope.ai/collections/Qwen/Qwen38-Flash-Next
Source: antirez/ds4