YZ Index
评测数据
当前展示:Run #363 | 2026-10-06 | 128 题库 | 公式 v7 | 判分 v6.4
数据公开说明:为防止题库污染和过拟合,题目原文和预期答案不公开。本页展示模型回答、得分、判分方式等透明数据。完整方法论请参阅方法论页面。
| 模型 | 代码执行 | 材料约束 | 工程判断 | 任务表达 | 诚信 | 主榜分 | 性价比 | 稳定性 | 可用性 | 逐题 |
|---|---|---|---|---|---|---|---|---|---|---|
| GPT-6 Luna gpt | 75.00 | 88.30 | 100.00 | 100.00 | 70.00 pass | 80.99 | 100.0 | 50.3 | 100.0 | |
| GPT-6 Sol gpt | 75.00 | 88.30 | 94.80 | 100.00 | 70.00 pass | 80.99 | 96.6 | 50.6 | 100.0 | |
| GPT-5.5 gpt | 75.00 | 81.70 | 100.00 | 100.00 | 75.00 pass | 78.02 | 74.6 | 37.5 | 100.0 | |
| GPT-o3 gpt | 75.00 | 81.70 | 100.00 | 63.90 | 80.00 pass | 78.02 | 43.2 | 46.2 | 95.0 | |
| Grok 4 grok | 73.70 | 81.70 | 66.70 | 61.10 | 55.00 warn | 77.30 | 82.4 | 25.5 | 100.0 | |
| 豆包 Pro doubao | 75.00 | 68.30 | 100.00 | 100.00 | 65.00 pass | 71.99 | 100.0 | 28.5 | 100.0 | |
| Claude Sonnet 4.6 claude | 50.00 | 93.30 | 88.90 | 75.00 | 70.00 pass | 69.49 | 85.9 | 28.7 | 100.0 | |
| Claude Opus 4.7 claude | 50.00 | 88.30 | 100.00 | 58.90 | 100.00 pass | 67.24 | 25.5 | 32.7 | 100.0 | |
| GPT-6 Astra gpt | 50.00 | 88.30 | 97.80 | 100.00 | 80.00 pass | 67.24 | 34.3 | 38.7 | 100.0 | |
| Qwen3 Max qwen | 46.00 | 88.30 | 41.80 | 28.90 | 70.00 pass | 65.04 | 98.4 | 26.1 | 100.0 | |
| GPT-6.1 Sol gpt | 50.00 | 63.30 | 97.00 | 100.00 | 80.00 pass | 55.99 | 91.8 | 29.5 | 100.0 | |
| DeepSeek V4 Pro deepseek | 41.70 | 63.30 | 100.00 | 100.00 | 90.00 pass | 51.42 | 94.9 | 50.3 | 90.0 | |
| Gemini 2.5 Pro gemini | 41.70 | 63.30 | 75.00 | 85.00 | 75.00 pass | 51.42 | 90.0 | 33.1 | 85.0 | |
| Gemini 3.1 Pro gemini | 41.70 | 63.30 | 75.00 | 75.00 | 70.00 pass | 51.42 | 72.6 | 19.4 | 100.0 | |
| GLM-4.6 zhipu | 0.00 | 20.00 | 25.00 | 75.00 | 100.00 pass | 9.00 | 67.2 | 23.7 | 55.0 |
API 访问:如需程序化访问评测数据,请使用我们的 API。