LongCat-Flash-Thinking-2601

model paper
Meituan
Updated 560B MoE agentic reasoning model with DORA RL framework and Heavy Thinking mode. Achieves 97.1% on theorem proving benchmarks. A distinct release from the September 2025 Flash-Thinking, with a new technical report.
Paper (arXiv)HuggingFaceGitHub

Outputs 2

LongCat-Flash-Thinking-2601

model
HuggingFaceGitHub
Architecture MOE
Parameters 560B
Active params 27B

LongCat-Flash-Thinking-2601 Technical Report

paper
Paper (arXiv)

arXiv: 2601.16725

reasoningmoetrainingagenticopen-weight

Related