OpenCompass LLM - Knowledge - English (CompassBench 2510): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 (2025-04-16) (High) | 93.8 |
| 2 | GPT-5 (Thinking) | 93.5 |
| 3 | Grok 4 | 91.8 |
| 4 | DeepSeek R1 0528 | 91 |
| 5 | O4 Mini (2025-04-16) (High) | 91 |
| 6 | Gemini 2.5 Pro | 90.5 |
| 7 | DeepSeek V3.2 Exp | 90.2 |
| 8 | Ling-1T | 90 |
| 9 | Claude Sonnet 4.5 (Thinking) | 89.8 |
| 10 | GLM-4.6 | 89.5 |
| 11 | Qwen 3 235B A22B 2507 (Thinking) | 89.2 |
| 12 | Kimi K2 0905 | 89 |
| 13 | Llama 4 Maverick Instruct | 89 |
| 14 | Hunyuan-T1-20250822 | 88.5 |
| 15 | Qwen 3 Next 80B A3B (Thinking) | 88.2 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-knowledge-english-compassbench-2510 · How It Works · Data refreshed daily, snapshot 2026-09-19.