The 2026-11-10 YZ Index Smoke quick test covered 13 models, with GPT-6 Astra and GPT-6.1 Sol tied for first at 92.49 points. Smoke is a daily 10-question quick test suited to observing short-term signals, not equivalent to Full weekly ranking conclusions.
This Smoke evaluation covered only two main leaderboard dimensions: Code Execution and Material Constraints. The main leaderboard formula is 0.55 × Code Execution + 0.45 × Material Constraints. Because the daily sample size is small, single-day scores are better used as monitoring signals rather than long-term conclusions about model capability.
Daily Rankings
| Rank | Model | Main Leaderboard | Code Execution | Material Constraints | Integrity |
|---|---|---|---|---|---|
| #1 | GPT-6 Astra | 92.49 | 100 | 83.3 | pass |
| #2 | GPT-6.1 Sol | 92.49 | 100 | 83.3 | pass |
| #3 | GPT-6 Luna | 87.42 | 81.7 | 94.4 | pass |
| #4 | Gemini 3.1 Pro | 82.42 | 81.7 | 83.3 | pass |
| #5 | GPT-5.5 | 79.93 | 82.4 | 76.9 | pass |
| #6 | GPT-6 Sol | 76.98 | 81.7 | 71.2 | pass |
| #7 | Doubao Pro | 73.19 | 69.5 | 77.7 | pass |
| #8 | Claude Opus 4.7 | 66.37 | 57.1 | 77.7 | pass |
| #9 | Gemini 2.5 Pro | 65.02 | 48.5 | 85.2 | pass |
| #10 | Claude Sonnet 4.6 | 64.89 | 73.3 | 54.6 | pass |
| #11 | GPT-o3 | 63.99 | 44.1 | 88.3 | pass |
| #12 | Grok 4 | 61.84 | 37.4 | 91.7 | pass |
| #13 | DeepSeek V4 Pro | 58.75 | 25 | 100 | pass |
Data Interpretation
Today’s top two on the main leaderboard, GPT-6 Astra and GPT-6.1 Sol, both scored 92.49, with Code Execution at 100 and Material Constraints at 83.3, showing a structure of full marks in Code Execution paired with 83.3 in Material Constraints. Third place GPT-6 Luna scored 87.42 on the main leaderboard, with Code Execution at 81.7 and Material Constraints at 94.4; its Material Constraints score was higher than the top two, while its Code Execution was relatively lower. Fourth place Gemini 3.1 Pro scored 82.42 on the main leaderboard, with Code Execution at 81.7 and Material Constraints at 83.3, forming a different strength profile from the models above.
Grok 4 fell 29.5 points on the main leaderboard, with Code Execution down 56.4 points. GPT-o3 fell 28.5 points on the main leaderboard, with Code Execution down 55.9 points and Material Constraints up 5 points. Claude Sonnet 4.6 fell 26.1 points on the main leaderboard, with Code Execution down 26.7 points and Material Constraints down 25.4 points. Gemini 2.5 Pro fell 17.5 points on the main leaderboard, with Code Execution down 26.5 points and Material Constraints down 6.5 points. GPT-6 Sol fell 15.5 points on the main leaderboard, with Code Execution down 18.3 points and Material Constraints down 12.1 points. These score changes may come from question sampling fluctuation or may be real degradation, and require follow-up run review to confirm.
Smoke is a small-sample single-day signal; the above observations reflect only the data structure for that day and do not make long-term inferences.
Key Changes
- Grok 4: Main leaderboard down 29.5 points, Code Execution -56.4 points
- GPT-o3: Main leaderboard down 28.5 points, Code Execution -55.9 points, Material Constraints +5 points
- Claude Sonnet 4.6: Main leaderboard down 26.1 points, Code Execution -26.7 points, Material Constraints -25.4 points
- Gemini 2.5 Pro: Main leaderboard down 17.5 points, Code Execution -26.5 points, Material Constraints -6.5 points
- GPT-6 Sol: Main leaderboard down 15.5 points, Code Execution -18.3 points, Material Constraints -12.1 points
Signals to Watch
- No publishable anomaly signals were retained in this edition.
When reading this type of Smoke brief, focus on two questions: first, whether a model has exposed the same type of weakness for several consecutive days; second, whether its Integrity rating has moved from pass to warn or fail. Large single-day changes in execution or constraint scores may come from question sampling or may be early signals of real degradation, and require follow-up run review.
Data source: YZ Index | Run #420 | View raw data
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接