YZ Index
Evaluation Data
Currently showing：Run #87 | 2026-04-27 | 212-question pool | Formula v7 | Judge set v6
Data Disclosure：To prevent benchmark contamination and overfitting, question texts and expected answers are not disclosed. This page shows model responses, scores, and judging methods for transparency. For the full methodology, seeMethodology page。
| Model | Code Execution | Grounding | Engineering Judgment | Task Communication | Integrity | Overallpts | Value | Stability | Availability | Per Task |
|---|---|---|---|---|---|---|---|---|---|---|
| Grok 3 grok | 88.90 | 84.40 | 43.50 | 40.00 | 77.50 pass | 86.88 | 25.8 | 35.5 | 99.0 | |
| 豆包 Pro doubao | 92.20 | 79.40 | 46.30 | 40.00 | 77.50 pass | 86.44 | 93.3 | 38.8 | 100.0 | |
| Gemini 2.5 Pro gemini | 89.40 | 78.10 | 47.20 | 40.00 | 80.80 pass | 84.32 | 39.3 | 37.7 | 100.0 | |
| Claude Sonnet 4.6 claude | 86.50 | 81.10 | 43.80 | 40.00 | 74.20 pass | 84.07 | 25.1 | 35.7 | 99.0 | |
| Claude Opus 4.6 claude | 86.50 | 79.70 | 46.30 | 40.00 | 67.50 pass | 83.44 | 5.1 | 35.2 | 100.0 | |
| DeepSeek V3 deepseek | 83.20 | 77.80 | 44.30 | 40.00 | 59.20 warn | 80.77 | 99.7 | 32.8 | 100.0 | |
| Qwen Max qwen | 78.40 | 77.30 | 40.70 | 40.00 | 65.80 pass | 77.91 | 48.6 | 32.7 | 100.0 | |
| DeepSeek R1 deepseek | 78.90 | 72.20 | 38.70 | 40.00 | 54.20 warn | 75.89 | 90.3 | 30.2 | 100.0 | |
| 文心一言 4.0 ernie | 77.00 | 72.30 | 39.70 | 40.00 | 69.20 pass | 74.89 | 98.6 | 31.3 | 100.0 | |
| GPT-4o gpt | 71.70 | 57.60 | 41.50 | 40.00 | 74.20 pass | 65.36 | 29.1 | 30.4 | 91.0 | |
| GPT-o3 gpt | 73.40 | 49.20 | 38.70 | 40.00 | 69.20 pass | 62.51 | 7.0 | 28.9 | 87.0 |
API Access：For programmatic access to evaluation data, please use our API。