On 2026-11-09, the YZ Index Smoke quick test covered 13 models, with GPT-6 Sol and GPT-o3 tying for first place of the day at 92.49 points. Smoke is a daily 10-question quick test, suited to observing short-term signals; it is not equivalent to conclusions from the Full weekly leaderboard.
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 as long-term verdicts on model capability.
Rankings for the Day
| Rank | Model | Main | Code Execution | Material Constraints | Integrity |
|---|---|---|---|---|---|
| #1 | GPT-6 Sol | 92.49 | 100 | 83.3 | pass |
| #2 | GPT-o3 | 92.49 | 100 | 83.3 | pass |
| #3 | Grok 4 | 91.33 | 93.8 | 88.3 | pass |
| #4 | Claude Sonnet 4.6 | 91 | 100 | 80 | pass |
| #5 | GPT-6 Luna | 90.24 | 100 | 78.3 | pass |
| #6 | GPT-6 Astra | 87.99 | 100 | 73.3 | pass |
| #7 | GPT-6.1 Sol | 87.99 | 100 | 73.3 | pass |
| #8 | Doubao Pro | 83.49 | 100 | 63.3 | pass |
| #9 | Gemini 2.5 Pro | 82.52 | 75 | 91.7 | pass |
| #10 | Claude Opus 4.7 | 78.74 | 75 | 83.3 | pass |
| #11 | Gemini 3.1 Pro | 78.74 | 75 | 83.3 | pass |
| #12 | GPT-5.5 | 78.74 | 75 | 83.3 | pass |
| #13 | DeepSeek V4 Pro | 68 | 50 | 90 | pass |
Data Interpretation
Today's Smoke quick test shows GPT-6 Sol and GPT-o3 with identical main leaderboard scores of 92.49, both at 100 on code execution and 83.3 on material constraints, a combination that places them at the front. Grok 4 scores 91.33 on the main leaderboard, with code execution at 93.8 and material constraints at 88.3, its material constraint score being relatively outstanding. Claude Sonnet 4.6 has code execution at 100 and material constraints at 80, for a main leaderboard score of 91. The leading models mostly show strong code execution but vary in material constraints.
Gemini 2.5 Pro rose 30.5 points on the main leaderboard versus the previous run, with code execution up 55 points; GPT-o3 rose 29.1 points, with code execution up 55.6 points and integrity turning to pass; Doubao Pro rose 23.9 points, with code execution up 46.7 points; Claude Opus 4.7 rose 23.4 points, with code execution up 49 points but material constraints down 7.8 points. These movers all show notable changes in the code execution dimension.
GLM-4.6 and Qwen3 Max did not enter the rankings because of incomplete data caused by API failures; this may be sampling fluctuation or genuine degradation, and follow-up runs are needed to confirm. As a small-sample single-day signal, changes in Smoke quick-test scores should be interpreted with restraint.
Key Changes
- Gemini 2.5 Pro: main leaderboard up 30.5 points, code execution +55 points
- GPT-o3: main leaderboard up 29.1 points, code execution +55.6 points, integrity warn→pass
- Doubao Pro: main leaderboard up 23.9 points, code execution +46.7 points, integrity warn→pass
- Claude Opus 4.7: main leaderboard up 23.4 points, code execution +49 points, material constraints −7.8 points
- Gemini 3.1 Pro: main leaderboard up 22.6 points, code execution +49.2 points, material constraints −10 points
Signals to Watch
- GLM-4.6: incomplete data (missing communication dimension, API failure/timeout); has entered the automatic re-run queue and is not ranked this period
- Qwen3 Max: incomplete data (API failure/timeout); has entered the automatic re-run queue and is not ranked this period
When reading this kind of Smoke brief, the focus should be on two questions: first, whether a model has exposed the same type of weakness on consecutive days; second, whether its integrity rating has moved from pass into warn or fail. Large single-day changes in execution or constraint scores may stem from question sampling, or may be an early signal of genuine degradation, and require follow-up runs to confirm.
Data source: YZ Index | Run #418 | View raw data
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接