Qwen3.8-Flash-Next SGLang Deployment
Qwen4's 51B N-gram embeddings scale model capacity with minimal extra compute per token, making it efficient for long-horizon tasks in the region.
Latest
Every published item, in order · 46 items
Aug 26
Alibaba releases Qwen3.8-Flash-Next
Cost-efficient alternative to Claude Opus 4.6, potentially more accessible for ASEAN developers and businesses.
Chinese contentQwen3.8-Flash 正式发布，Qwen3.8-Flash-Next 正式开源！
Qwen3.8-Flash-Next 开源，效率提升显著，成本降低，性能优异，是 Qwen4 架构的早期预览，值得尝试！
Qwen3.8-Flash-Next-Base model tops benchmarks
Qwen3.8-Flash-Next-Base offers competitive performance with fewer parameters, potentially reducing computational costs for users in ASEAN and China.
Aug 25
Chinese content阿里千问 Qwen3.8-Flash-Next 开源预告
阿里千问发布新模型，开源 Qwen3.8-Flash-Next，多模态 MoE 架构，为 Qwen4 模型家族做准备，值得一试！
IT之家Source
