OpenCompass Reasoning - Abductive - Chinese (CompassBench 2404): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 31 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4 Turbo (Preview) | 88.8 |
| 2 | OrionStar-Yi-34B-Chat | 77.5 |
| 3 | Yi 34B (Chat) | 69.4 |
| 4 | WizardLM-70B-V1.0 | 60.2 |
| 5 | deepseek-llm-7B-chat | 51 |
| 6 | Qwen-14B-Chat | 39.8 |
| 7 | Baichuan2-7B-Chat | 34.7 |
| 8 | WizardLM-13B-V1.2 | 31.6 |
| 9 | Qwen-72B-Chat | 29.6 |
| 10 | Llama 2 13B Chat | 28.6 |
| 11 | Mistral 7B Instruct (v0.2) | 22.4 |
| 12 | Qwen-7B Chat | 22.4 |
| 13 | Llama 2 7B Chat | 16.3 |
| 14 | Baichuan2-13B-Chat | 11.2 |
| 15 | Llama 2 70B Chat | 10.2 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-reasoning-abductive-chinese-compassbench-2404 · How It Works · Data refreshed daily, snapshot 2026-09-19.