Meituan releases efficient inference model LongCat-Flash-Thinking

2025-09-22 14:12:31

On September 22nd, Meituan released LongCat-Flash-Thinking, a highly efficient inference model. Meituan stated that based on AIME25 test data, LongCat-Flash-Thinking demonstrated more efficient agent tool invocation within the framework, saving 64.5% of tokens compared to no tool invocation while maintaining 90% accuracy. LongCat-Flash-Thinking is now open source on HuggingFace and Github, and is available for trial on its official website.

AI Meituan

Email Subscription

Newsletters and emails are now available! Delivered on time, every weekday, to keep you up to date with North American business news.

Weekly Highlights

                                    Musk: XAI will certainly raise funds in the next few months, but not now
                            2025-09-20

                                    Nvidia (NVDA.O) fell further, dropping nearly 3%.
                            2025-09-17

                                    SoftBank and OpenAI's AI joint venture in Japan is severely delayed, with an update now expected in November.
                            2025-09-18

                                    Alipay's two operating entities have completed the name change
                            2025-09-19

                                    Hande Information: The company has cooperation with Oracle
                            2025-09-23