[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-0c28281d-1b70-4206-ae2d-3219f1e4fbb9":3},{"id":4,"title":5,"summary":6,"original_url":7,"source_id":8,"tags":9,"published_at":23,"created_at":24,"modified_at":25,"is_published":26,"publish_type":27,"image_url":13,"view_count":28},"0c28281d-1b70-4206-ae2d-3219f1e4fbb9","Qwen3-Coder-Next：稀疏MoE架构重塑代码智能效率边界","5月8日，阿里Qwen团队发布Qwen3-Coder-Next，一款专为主动式编程（Agentic Coding）设计的80B参数MoE模型，每次前向传播仅激活3B参数，却具备与Dense模型相当的编程能力，引发开放权重社区广泛讨论。\n\n核心技术在于Gated DeltaNet配合Gated Attention，将Attention的二次计算复杂度转为线性，使模型得以在维持262K token超长上下文的同时避免指数级延迟惩罚。在仓库级任务中，吞吐量比同级别Dense模型提升约10倍。训练阶段引入Best-Fit Packing策略，有效缓解了长上下文场景下的幻觉问题，保持了上下文信息的完整性。\n\n该模型以Apache 2.0许可证开源，权重已在HuggingFace发布4个变体，并附有详细技术报告。在编程Agent成为行业竞争焦点的当下，小激活、大能力的稀疏MoE设计为本地部署提供了全新范式——开发者得以在消费级硬件上，以3B模型的资源消耗，获得80B量级的结构化代码理解能力，直接冲击了此前只有闭源大模型才能触及的能力天花板。","https:\u002F\u002Fventurebeat.com\u002Ftechnology\u002Fqwen3-coder-next-offers-vibe-coders-a-powerful-open-source-ultra-sparse","17ff6400-4413-4b16-86fb-99951dbbd08d",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"e676a5cf-1f24-472f-a765-86fa21a1bc3c","ai-model",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"e82b2d09-81b2-43d1-977e-e018443b3c14","coding-agent",{"id":18,"name":19,"slug":19,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":21,"name":22,"slug":22,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source","2026-05-08T10:00:00Z","2026-05-08T10:05:10.079600Z","2026-05-08T10:05:10.079611Z",true,"agent",4]