OpenCompass Knowledge - Science - English (CompassBench 2409): leaderboard
OpenCompass CompassBench 2409 edition (file dated 2024-10-09): science sub-task of the knowledge category, English questions. CompassBench is OpenCompass's periodic evaluation of chat models on a private question set; each edition is a separate board. Score 0-100, higher is better.
Metric: Score (%). Source: rank.opencompass.org.cn. Saturation forecast: Estimated already saturated. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o (2024-05-13) | 94.4 | #337 |
| 2 | Llama 3.1 405B Instruct FP8 | 92 | #400 |
| 3 | Step 2 16K | 92 | #442 |
| 4 | Claude 3.5 Sonnet (20240620) | 91.2 | #308 |
| 5 | GPT-4o Mini (2024-07-18) | 91.2 | #568 |
| 6 | GPT-4o (2024-08-06) | 91.2 | #343 |
| 7 | Qwen 2.5 72B Instruct | 90.4 | #469 |
| 8 | GLM-4 Plus | 89.6 | #398 |
| 9 | Llama 3.1 70B Instruct | 88.8 | #567 |
| 10 | Mistral Large 2 (Jul) | 88.8 | #490 |
| 11 | Yi Large | 86.4 | #566 |
| 12 | ERNIE-4.0-Turbo-8K-Preview | 84.8 | #603 |
| 13 | Qwen 2.5 7B Instruct | 84 | #863 |
| 14 | Gemma 2 9B (IT) | 81.6 | #795 |
| 15 | Llama 3.1 8B Instruct | 80.8 | #1017 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-knowledge-science-english-compassbench-2409 · How It Works · Data refreshed daily, snapshot 2026-11-05.