LongCat-2.5: benchmark results
Provider: Meituan. Released 2026-11-02. Access: API.
Aggregate Rating 1829 ± 38, rank #50 of 10916 rated models, from 26 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LongCat 2.5 Release - AutomationBench - v1.0.6 | 51.8 | Score (%) | 100 |
| LongCat 2.5 Release - BrowseComp | 91.9 | Score (%) | 100 |
| LongCat 2.5 Release - CharXiv-RQ - w/ tools | 90.2 | Score (%) | 100 |
| LongCat 2.5 Release - DeepSWE - v1.1 | 64.6 | Score (%) | 100 |
| LongCat 2.5 Release - HLE - w/ tools | 63.1 | Score (%) | 100 |
| LongCat 2.5 Release - JobBench | 65.4 | Score (%) | 100 |
| LongCat 2.5 Release - NL2Repo | 60.5 | Score (%) | 100 |
| LongCat 2.5 Release - Program Bench - Almost Resolved | 25.5 | Score (%) | 100 |
| LongCat 2.5 Release - ResearchRubrics | 81 | Score (%) | 100 |
| LongCat 2.5 Release - SkillsBench | 63.3 | Score (%) | 100 |
| LongCat 2.5 Release - Terminal Bench v2.1 | 88.4 | Score (%) | 100 |
| LongCat 2.5 Release - Terminal Bench v4.0 | 38.5 | Score (%) | 100 |
Interactive version: theaggregate.ai/model?slug=longcat-2-5 · How It Works · Data refreshed daily, snapshot 2026-11-08.