Meituan
What its model cards measure
Tall ticks are Meituan model cards that list the benchmark in their headline table. Gray dots are cards that leave it out. Rows are the benchmarks the lab listed most, at least twice each, in the order it started using them.
GPQA Diamond IFEval MMLU-Pro MATH-500 AIME 2024 AIME 2025 Beyond AIME ZebraLogic LiveCodeBench SWE-bench Verified tau2-bench VITA-Bench MMLU Arena-Hard v2.0 C-Eval CMMLU COLLIE Meeseeks DROP GraphWalks HumanEval+ MBPP+ LongCat safety evaluation (Meituan) HMMT 2025 OJBench IMO-AnswerBench BrowseComp RWSearch
Model cards
| Date | Model | Document | Benchmarks |
|---|---|---|---|
| 5 Jul 2026 | LongCat-2.0 | LongCat-2.0 model card (Hugging Face) | 10 |
| 14 Jan 2026 | LongCat-Flash-Thinking-2601 | LongCat-Flash-Thinking-2601 model card (Hugging Face) | 18 |
| 31 Oct 2025 | LongCat-Flash-Omni | LongCat-Flash-Omni model card (Hugging Face) | 57 |
| 21 Sep 2025 | LongCat-Flash-Thinking | LongCat-Flash-Thinking model card (Hugging Face) | 21 |
| 29 Aug 2025 | LongCat-Flash-Chat | LongCat-Flash-Chat model card (Hugging Face) | 25 |
Every Meituan model
- LongCat-2.0 Jul 2026
- LongCat-Flash-Thinking-2601 Jan 2026
- LongCat-Flash-Omni Oct 2025
- LongCat-Flash-Thinking Sep 2025
- LongCat-Flash-Chat Aug 2025