[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-kimi-k2-6-trillion-moe-1t-32b-active-256k":3,"news-related-d4d40e4b-04e9-45cf-a04c-792aca45b152":34},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":26,"news_slug":27,"published_at":28,"created_at":29,"modified_at":30,"is_published":31,"publish_type":32,"image_url":13,"view_count":33},"d4d40e4b-04e9-45cf-a04c-792aca45b152","Kimi K2.6开源发布：万亿参数MoE模型的长时编程与Agent Swarm突破","\nMoonshot AI于4月20日开源发布Kimi K2.6，这是一款万亿参数级别的原生多模态MoE模型，在长时编程和大规模Agent协作方面展现出显著的工程突破。\n\nK2.6采用1万亿总参数、每Token仅激活320亿参数的MoE架构，包含384个专家模块，每次推理选择8个专家加1个共享专家。模型使用多头潜在注意力（MLA）机制，上下文窗口达256K tokens，视觉能力通过4亿参数的MoonViT编码器原生集成，而非后期拼接。\n\n在编程基准测试中，K2.6表现亮眼。SWE-Bench Pro得分58.6，超越GPT-5.4（57.7）和Claude Opus 4.6（53.4）；在被称为\"最难知识基准\"的Humanity's Last Exam（工具版）中，K2.6以54.0分领先所有对比模型，包括GPT-5.4和Claude Opus 4.6。\n\n最具工程价值的突破在于长时自主编程能力。在13小时的连续执行中，K2.6自主重构了一个8年历史的金融撮合引擎，通过12轮优化策略迭代，发起超过1000次工具调用，修改4000多行代码，最终将中等吞吐量提升185%。另一个案例中，模型在Zig语言中实现了Qwen3.5-0.8B的本地推理优化，4000多次工具调用后将吞吐量从15提升至193 tokens\u002Fsec。\n\nAgent Swarm能力同样值得关注。K2.6支持最多300个子Agent并行执行、4000个协调步骤同时运行，通过水平扩展而非垂直加深推理链来解决复杂任务。这种大规模并行Agent架构为实际工程部署提供了新的范式。\n\nK2.6以Modified MIT许可证开源，权重已在Hugging Face发布，可通过vLLM、SGLang或KTransformers部署。","https:\u002F\u002Fkimi-k2.org\u002Fblog\u002F24-kimi-k2-6-release","5e4fd3d1-9cb4-44a6-bae5-9ffb449c05c1",[10,14,17,20,23],{"id":11,"name":12,"slug":12,"description":13,"color":13},"a8002d98-9df1-4ab9-94d4-a7625af634c4","china-ai",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},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":21,"name":22,"slug":22,"description":13,"color":13},"b1853a5a-d940-42b7-94f9-0488ee3f2cf7","new-model",{"id":24,"name":25,"slug":25,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[],"kimi-k2-6-trillion-moe-1t-32b-active-256k","2026-04-24T03:00:00Z","2026-04-23T22:09:06.555175Z","2026-08-19T02:08:40.142862Z",true,"agent",160,{"items":35},[36,41,46,51,56,61],{"id":37,"title":38,"news_slug":39,"published_at":40},"fc0cd2d4-cc5f-4f97-b91f-08719e41e8ec","Qwen3.6-27B：27B密集模型超越397B MoE，单卡部署的编程新选择","qwen-3-6-27b-dense-beats-397b-moe-coding-77pct","2026-04-24T03:30:00+00:00",{"id":42,"title":43,"news_slug":44,"published_at":45},"49d0aaf8-6fcf-4bf9-83fb-19a016ae2784","CompactionRL:把上下文压缩塞进 RL 循环,GLM-5.2 训练管线吃下 5–7pp 编码代理增益","compactionrl-glm-5-2","2026-07-10T12:08:00+00:00",{"id":47,"title":48,"news_slug":49,"published_at":50},"fa8a0c53-dd89-4870-8e6e-978ce038a919","GitHub Copilot 上线 Kimi K2.7：开源权重模型首次进入默认模型选择器","github-copilot-kimi-k2-7","2026-07-03T06:00:00+00:00",{"id":52,"title":53,"news_slug":54,"published_at":55},"61222f5d-b7e6-4200-8517-3b1972040d24","小米开源 MiMo Code：把 Coding Agent 拆成「计算-记忆-演化」三段式，长程编程首次跑通","xiaomi-mimo-code-compute-memory-evolution","2026-06-11T04:00:00+00:00",{"id":57,"title":58,"news_slug":59,"published_at":60},"88d40bcd-f92f-481e-b644-5da3dd9813a4","四家中国实验室十二天内密集发布开源代码模型：前沿能力与低成本并行","4-china-labs-12-days-coding-open","2026-05-21T02:30:00+00:00",{"id":62,"title":63,"news_slug":64,"published_at":65},"010979e2-4e0a-4dcc-8a91-c6a99bfebc04","腾讯混元 Hy3 Preview 开源：295B MoE 剑指 Agent 实用性","tencent-hunyuan-hy3-preview-295b-moe","2026-05-08T13:00:00+00:00"]