Meituan LongCat releases LongCat-2.5 native multimodal model, 1.6T parameters targeting long-horizon agent tasks
English text is being prepared.
Meituan LongCat releases the native multimodal model LongCat-2.5, with a total parameter count of 1.6 trillion and approximately 48 billion parameters activated per token, strengthening native visual understanding and agentic coding capabilities on the basis of LongCat-2.0. The model incorporates more long-horizon trajectory data into long-sequence training, and through large-scale long-horizon reinforcement learning enhances end-to-end engineering capabilities such as complex codebase development and complete project construction, while representing visual signals as multi-level discrete tokens to uniformly process visual and textual inputs. The official statement says it has optimized long-context training and W4A4 inference on domestic compute, and it will soon be available on GitHub and HuggingFace.
Why read this
AI score 68
The original text provides the 1.6T parameter and activation scale, and explains the specific approach of the long-horizon agent training pipeline, from which its engineering capability orientation can be judged.
Topics
Body
美团龙猫发布 LongCat-2.5 原生多模态 AI 模型:1.6T 参数,面向长程智能体任务
2026-11-02 09:18 IT之家 - 沁沧(实习)
感谢IT之家网友 GreatMOLA 的线索投递!
IT之家 11 月 2 日消息,美团龙猫今日发布面向长程智能体任务的原生多模态模型 LongCat-2.5,模型总参数量为 1.6 万亿,每个 token 激活约 480 亿参数,在 LongCat-2.0 的基础上进一步强化原生视觉理解与智能体编程能力。
据介绍,龙猫在长序列训练中加入更多长程轨迹数据,并通过大规模长程强化学习,提升模型在复杂代码库开发、完整项目构建等任务中的端到端工程能力。围绕数据合成、环境构建与跨 Harness 混合训练,官方持续扩展这条训练链路。此外,龙猫将真实 Agent 的失败轨迹转化为可复现、可验证的训练任务,针对模型能力短板持续补充训练数据。
与此同时,LongCat-2.5 将视觉信号表示为多层级离散 token,使主干模型以统一的 token 形式处理视觉与文本输入。原生视觉理解与智能体编程能力的结合,使模型在 Visual Coding 上表现出色,并支持创意设计和 3D 等任务。
官方表示,这些升级由持续演进的基础设施支撑:优化国产算力上的长上下文训练与 W4A4 推理,并通过训推数值一致性和长上下文 rollout 优化,支撑大规模长程 RL 的高效运行。
IT之家附评测结果:
Github、HuggingFace:即将上线
广告声明:文内含有的对外跳转链接(包括不限于超链接、二维码、口令等形式),用于传递更多信息,节省甄选时间,结果仅供参考,IT之家包含外链的文章均包含本声明。