OpenCompass Reasoning - Social - English (CompassBench 2409): leaderboard
OpenCompass CompassBench 2409 edition (file dated 2024-10-09): social reasoning sub-task of the reasoning 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 | Claude 3.5 Sonnet (20240620) | 63.8 | #308 |
| 2 | Qwen 2.5 72B Instruct | 60.2 | #469 |
| 3 | Step 2 16K | 58.9 | #442 |
| 4 | Gemini 1.5 Pro | 57.3 | #454 |
| 5 | Mistral Nemo Instruct (2407) | 57.3 | #917 |
| 6 | GLM-4 Plus | 57.3 | #398 |
| 7 | Gemma 2 9B (IT) | 56.5 | #795 |
| 8 | Gemma 2 27B (IT) | 55 | #652 |
| 9 | Mistral Large 2 (Jul) | 54.4 | #490 |
| 10 | Llama 3.1 405B Instruct FP8 | 53.2 | #400 |
| 11 | Mistral-Small-Instruct-2409 | 52.6 | #709 |
| 12 | Llama 3.1 70B Instruct | 51.8 | #567 |
| 13 | GPT-4o (2024-05-13) | 49.7 | #337 |
| 14 | Qwen 2.5 7B Instruct | 48.4 | #863 |
| 15 | Yi-1.5-9B Chat | 46.9 | #1032 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-reasoning-social-english-compassbench-2409 · How It Works · Data refreshed daily, snapshot 2026-11-05.