美团LongCat发布重思考模式总结模型
美团LongCat推出基于5600亿参数MoE架构大模型LongCat-Flash-Thinking-2601的重思考模式（Heavy Thinking Mode），并发布LongCat-HeavyModel-Summary模型。该模式通过并行思考与总结两阶段协同扩展推理能力：前者以高温度并行生成多路径扩展宽度，后者将精炼轨迹递归反馈形成迭代循环延伸深度。模型经额外强化学习优化总结能力，已在Longcat AI平台上线。
美团开源 560B 参数 MoE 推理模型，Heavy Thinking 模式支持并行多路径探索，已上线可体验
Model Introduction
We introduce an updated version of LongCat-Flash-Thinking, named LongCat-Flash-Thinking-2601, a powerful and efficient Large Reasoning Model (LRM) with 560 billion total parameters, built upon an innovative Mixture-of-Experts (MoE) architecture.
To push reasoning capability beyond current boundary, we established our Heavy Thinking Mode based on the LongCat-Flash-Thinking-2601. Specifically, we decompose challenging problem solving into two complementary stages: parallel thinking and summarization, thus jointly scaling both reasoning depth and width. For reasoning width scaling, under Heavy Thinking Mode, multiple trajectories are independently generated in a parallel manner, enabling broad exploration of reasoning paths. Reasonably high inference temperature here is applied to ensure possible diversity. For reasoning depth scaling, the refined trajectories during the summarization stage can be recursively fed back into the summary model, forming an iterative reasoning loop that supports progressively deeper reasoning. An additional reinforcement learning stage is specifically tailored to train the summarization ability, thus further unlocking the potential of this mode.
We now release our LongCat-HeavyModel-Summary model at link, which is further trained based on LongCat-Flash-Thinking-2601.
We've launched Heavy Thinking Mode on the Longcat AI platform. Feel free to try it out: https://longcat.chat/.
来源：美团 LongCat：HuggingFace 新模型 · huggingface.co