The 2026-11-04 YZ Index Smoke Quick Test covered 12 models, with GPT-5.5 ranking first for the day at 94.8 points. Smoke is a daily 10-question quick test, suitable for observing short-term signals and not equivalent to conclusions from the weekly Full ranking.
This Smoke evaluation covers 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 Ranking
| Rank | Model | Main | Code Execution | Material Constraints | Integrity |
|---|---|---|---|---|---|
| #1 | GPT-5.5 | 94.8 | 99.3 | 89.3 | pass |
| #2 | Claude Opus 4.7 | 93.46 | 88.1 | 100 | pass |
| #3 | GPT-o3 | 92.59 | 96.1 | 88.3 | pass |
| #4 | GPT-6 Sol | 89.43 | 99.1 | 77.6 | pass |
| #5 | GPT-6 Astra | 88.05 | 96.6 | 77.6 | pass |
| #6 | GPT-6.1 Sol | 88.05 | 96.6 | 77.6 | pass |
| #7 | GPT-6 Luna | 75.79 | 74.3 | 77.6 | warn |
| #8 | Claude Sonnet 4.6 | 73.76 | 78.8 | 67.6 | warn |
| #9 | Grok 4 | 66.11 | 56.7 | 77.6 | pass |
| #10 | Doubao Pro | 56.73 | 60.1 | 52.6 | pass |
| #11 | Gemini 3.1 Pro | 40.61 | 30.8 | 52.6 | pass |
| #12 | Gemini 2.5 Pro | 23.67 | 0 | 52.6 | pass |
Data Interpretation
In today's YZ Index Smoke Quick Test, leading models showed clear divergence in how Code Execution and Material Constraints paired. GPT-5.5 scored 99.3 in Code Execution and 89.3 in Material Constraints for a main leaderboard score of 94.8; Claude Opus 4.7 relied on 100 in Material Constraints and 88.1 in Code Execution to reach 93.46; GPT-o3's 96.1 in Code Execution and 88.3 in Material Constraints corresponded to 92.59 on the main leaderboard. All three passed the Integrity check. This complementary combination of strong Code Execution and high Material Constraints placed them at the top under the main leaderboard formula.
Several models showed significant shifts. Gemini 2.5 Pro fell 40 points on the main leaderboard and 70.8 points in Code Execution; Gemini 3.1 Pro fell 33.7 points on the main leaderboard and 66.1 points in Code Execution; GPT-o3 rose 17.8 points on the main leaderboard and 37.5 points in Material Constraints; Claude Opus 4.7 rose 17.5 points on the main leaderboard and 18.1 points in Code Execution. GPT-6 Luna plunged 11.5 points on the main leaderboard, Claude Sonnet 4.6 plunged 11.1 points, and Doubao Pro plunged 13.2 points. These changes may stem from question sampling fluctuations or may reflect real single-day performance degradation, and need confirmation through subsequent run review.
The Smoke Quick Test is a small-sample, single-day signal. Outlier models such as the Gemini series and GPT-6 Luna showed relatively large fluctuations, and the current data reflects only the day's status with no long-term trend observed. DeepSeek V4 Pro, GLM-4.6, and Qwen3 Max had incomplete data due to API failures, have entered automatic rerun, and are not included in this period's ranking.
Key Changes
- Gemini 2.5 Pro: main leaderboard down 40 points, Code Execution -70.8 points
- Gemini 3.1 Pro: main leaderboard down 33.7 points, Code Execution -66.1 points, Material Constraints +5.9 points
- GPT-o3: main leaderboard up 17.8 points, Material Constraints +37.5 points
- Claude Opus 4.7: main leaderboard up 17.5 points, Code Execution +18.1 points, Material Constraints +16.7 points
- Doubao Pro: main leaderboard down 13.2 points, Code Execution -28.8 points, Material Constraints +5.9 points
Signals to Watch
- GPT-6 Luna: main leaderboard plunged -11.5 points
- Claude Sonnet 4.6: main leaderboard plunged -11.1 points
- Doubao Pro: main leaderboard plunged -13.2 points
- Gemini 3.1 Pro: main leaderboard plunged -33.7 points
- Gemini 2.5 Pro: main leaderboard plunged -40 points
- DeepSeek V4 Pro: data incomplete (missing execution dimension, API failure/timeout), has entered automatic rerun, and is not included in this period's ranking
- GLM-4.6: data incomplete (missing execution and communication dimensions, API failure/timeout), has entered automatic rerun, and is not included in this period's ranking
- Qwen3 Max: data incomplete (missing execution, judgment, integrity, and communication dimensions, API failure/timeout), has entered automatic rerun, and is not included in this period's ranking
When reading this kind of Smoke brief, focus on two questions: first, whether a given model has exposed the same type of weakness for multiple 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 subsequent run review.
Data source: YZ Index | Run #410 | View raw data
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接