【Tech 24H】On September 22, Meituan released the efficient inference model LongCat-Flash-Thinking. The new model has achieved top-tier performance among open-source models in public evaluations across multiple reasoning tasks, including logic, mathematics, coding and agent-based reasoning, with some tasks approaching the performance of closed-source models. The model has been open-sourced on Hugging Face and GitHub, with an online demo available.