OpenCompass LLM - Code (CompassBench 2507): leaderboard
OpenCompass CompassBench 2507 edition (file dated 2025-08-12): code category score. 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: Around February 2027. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 (2025-04-16) (High) | 50.6 | #125 (O3 (2025-04-16)) |
| 2 | O4 Mini (2025-04-16) (High) | 48.1 | #183 (O4 Mini (2025-04-16)) |
| 3 | GPT-5 | 47.9 | #105 |
| 4 | Grok 4 | 45.7 | #189 |
| 5 | Hunyuan-T1-20250711 | 43.8 | #277 |
| 6 | DeepSeek R1 0528 | 42.9 | #232 |
| 7 | Qwen 3 235B A22B 2507 (Thinking) | 41.6 | #263 |
| 8 | GLM 4.5 Air | 41.1 | #403 |
| 9 | Doubao-Seed-1.6-thinking-250715 | 40.4 | #169 |
| 10 | Gemini 2.5 Pro | 39.9 | #159 |
| 11 | GLM-4.5 | 39.6 | #283 |
| 12 | Claude Sonnet 4 (Thinking) | 38.2 | #207 (Claude Sonnet 4) |
| 13 | MiniMax M1 80k | 37.6 | #437 |
| 14 | DeepSeek V3 (0324) | 33.1 | #349 |
| 15 | Kimi K2 | 32.6 | #252 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-code-compassbench-2507 · How It Works · Data refreshed daily, snapshot 2026-11-05.