SciArena
Researchers ask scientific literature questions and vote between two models' long cited answers. Votes give Elo ratings.
- o3 1,151
- Claude Opus 4.1 1,126
- GPT-5 1,123
-
Gemini 3 Pro 1,086
- GPT-5.1 1,080
Official leaderboard
We track 2 benchmarks and 1 arena that Yale NLP made or helped make.
Top five in each of its 1 arena with three or more rated models, the most rated first.
Researchers ask scientific literature questions and vote between two models' long cited answers. Votes give Elo ratings.
Official leaderboard
| Name | Kind | Released | Best score | Status |
|---|---|---|---|---|
| MMVU 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. | Video understanding | Jan 2025 | | In use |
| BRIGHT-Pro Models retrieve passages that cover different parts of a question. The 739 questions include 115 from Earth Science Stack Exchange. | Questions that need search | Apr 2026 | | Not in cards |