OpenCompass Code - Comprehensive (CompassBench 2601): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 26 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 86 |
| 2 | Qwen 3 Max (2026-01-23) | 82.5 |
| 3 | GLM-4.7 | 79.5 |
| 4 | GPT-5.2 (High) | 79.5 |
| 5 | Kimi K2.5 | 79 |
| 6 | Claude Opus 4.5 (Thinking) | 78 |
| 7 | GPT-OSS-120B (High) | 77.5 |
| 8 | DeepSeek V3.2 | 76.5 |
| 9 | Kimi K2 (Thinking) | 74 |
| 10 | O4 Mini (2025-04-16) (High) | 74 |
| 11 | Grok 4.1 | 74 |
| 12 | MiniMax-M2.1 | 72 |
| 13 | Tencent HY 2.0 Think | 70.5 |
| 14 | Ring-1T | 66.5 |
| 15 | Qwen 3 235B A22B 2507 (Thinking) | 66 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-code-comprehensive-compassbench-2601 · How It Works · Data refreshed daily, snapshot 2026-09-19.