OpenCompass Reasoning - Common - English (CompassBench 2507): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4 (Thinking) | 76.7 |
| 2 | Qwen 3 235B A22B 2507 Instruct | 75.6 |
| 3 | Hunyuan-T1-20250711 | 75.6 |
| 4 | Doubao-Seed-1.6-thinking-250715 | 74.4 |
| 5 | GPT-5 | 73.3 |
| 6 | GLM-4.5 | 73.3 |
| 7 | GPT-4.1 (2025-04-14) | 70.9 |
| 8 | Kimi K2 | 69.8 |
| 9 | GLM 4.5 Air | 68.6 |
| 10 | MiniMax M1 80k | 68.6 |
| 11 | ERNIE-X1-Turbo-32K | 68.6 |
| 12 | DeepSeek V3 (0324) | 66.3 |
| 13 | MiniMax-Text-01 | 61.6 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-reasoning-common-english-compassbench-2507 · How It Works · Data refreshed daily, snapshot 2026-09-19.