LongCat-2.0
LongCat 2.0
LongCat-2.0 is a large language model released by Meituan, the Chinese food-delivery and local-services giant . In the Chinese model landscape it sits in the second tier: 葬AI's August 2026 benchmarking placed it at "mid-to-lower second-tier" among domestic models, roughly on par with DeepSeek V4 Pro and MiMo V2.5 Pro but trailing Hunyuan Hy3 and MiniMax M3 . Its earlier LongCat-Flash-Thinking variant, released in September 2025, was designed around a "fast, usable, balanced" philosophy with emphasis on agent capabilities and cost efficiency—Crossing noted its "dual-path reasoning framework" that lets the model decide when to compute internally versus call external tools, achieving AIME-25 accuracy near 90% at roughly 7,000 tokens per problem, below GPT-5 Thinking's consumption . Meituan has kept the model cheap (9.9 yuan for 50 million tokens) and does not force internal business units to adopt it .
AI-generated — may contain errors, please verify.
Coverage
Big Models Can Work, But Dongmodi Knows How to Fly
Which Browser Is Best for Running AI Models?
Meituan's LLM Is as Free as Its Delivery Service
Bullish on Meituan building world models
StepFun plays hardball after soft approach fails
Waiting for Yin to make his move.
Moonshot AI K3 Kicks Off a Lightning War on the Frontend
The big one is actually here.



