Back to BenchmarksFrontier
GSO-Bench
General science and observation benchmark
Leaderboard
(10 models)| Rank | Model | Score | Stderr |
|---|---|---|---|
| 1 | GPT-5.2 | 27.40 | — |
| 2 | Claude Opus 4.5 | 26.50 | — |
| 3 | Gemini 3 Pro | 18.60 | — |
| 4 | o3 | 8.80 | — |
| 5 | kimi-k2-thinking (official) | 4.90 | — |
| 6 | Qwen3-Max-Instruct | 4.90 | — |
| 7 | Claude 3.7 Sonnet | 4.60 | — |
| 8 | Gemini 2.5 Pro (Jun 2025) | 3.90 | — |
| 9 | o4-mini (high) | 3.60 | — |
| 10 | GPT-4o | 0.00 | — |
Data source: Epoch AI, “Data on AI Benchmarking”. Published at epoch.ai
Licensed under CC-BY 4.0