LongCat-Flash-Thinking

model paper
Meituan
Reasoning variant of LongCat-Flash trained via large-scale RL. Rivals proprietary systems in complex logic and STEM benchmarks.
Paper (arXiv)HuggingFaceGitHub

Outputs 2

LongCat-Flash-Thinking

model
HuggingFaceGitHub
Architecture MOE
Parameters 560B
Active params 27B

LongCat-Flash-Thinking: Scaling RL for Open Reasoning

paper
Paper (arXiv)

arXiv: 2509.18883

reasoningmoetrainingopen-weight

Related