On 2026-07-19, the YZ Index Smoke Quick Test covered 10 models, with Claude Opus 4.7 ranking first at 95.19. Smoke is a daily 10-question quick test suitable for monitoring short-term signals and is not equivalent to Full weekly ranking conclusions.
This Smoke evaluation only covers two main ranking dimensions: code execution and material constraints. The main ranking formula is 0.55 × Code Execution + 0.45 × Material Constraints. Due to the small daily sample size, single-day scores are more suitable as monitoring signals rather than long-term conclusions about model capabilities.
Daily Ranking
| Rank | Model | Main Score | Code Execution | Material Constraints | Integrity |
|---|---|---|---|---|---|
| #1 | Claude Opus 4.7 | 95.19 | 100 | 89.3 | pass |
| #2 | Claude Sonnet 4.6 | 90.02 | 99.2 | 78.8 | pass |
| #3 | Qwen3 Max | 82.23 | 96.9 | 64.3 | pass |
| #4 | GPT-5.5 | 81.44 | 75 | 89.3 | pass |
| #5 | Grok 4 | 79.73 | 71.9 | 89.3 | pass |
| #6 | Gemini 2.5 Pro | 76.49 | 74.6 | 78.8 | pass |
| #7 | GPT-o3 | 66.86 | 48.5 | 89.3 | pass |
| #8 | DeepSeek V4 Pro | 62.96 | 50 | 78.8 | pass |
| #9 | Gemini 3.1 Pro | 56.44 | 50 | 64.3 | pass |
| #10 | Doubao Pro | 49.21 | 25 | 78.8 | pass |
Data Interpretation
In today's YZ Index Smoke Quick Test, Claude Opus 4.7 ranks first with a main score of 95.19. Its combination of code execution 100 and material constraints 89.3 shows high performance in both areas; Claude Sonnet 4.6 follows closely with a main score of 90.02, with a structure of code execution 99.2 and material constraints 78.8 also leaning toward code execution; Qwen3 Max with a main score of 82.23 shows a notable gap between code execution 96.9 and material constraints 64.3; GPT-5.5 with a main score of 81.44 and Grok 4 with a main score of 79.73 both demonstrate strong support from material constraints at 89.3.
Compared to the previous run with the same scope, Grok 4's main score increased by 28.1 points, mainly from a 46.9-point rise in code execution; Claude Opus 4.7's main score rose by 16.1 points, with code execution up 25 points; Claude Sonnet 4.6's main score increased by 15.4 points, with code execution up 24.2 points; Qwen3 Max's main score rose by 12.1 points, with code execution up 32.2 points but material constraints down 12.5 points. In a small sample, these daily changes may result from question sampling fluctuations.
GPT-o3's main score plummeted by 13.8 points, and Gemini 3.1 Pro's material constraints dropped 26.6 points, requiring subsequent runs to verify whether these are real degradation signals; GLM-4.6 was not included in the ranking due to incomplete data. As a single-day signal, the above observations are described with restraint.
Key Changes
- Grok 4: main score +28.1, code execution +46.9, material constraints +5.2, integrity warn→pass
- Claude Opus 4.7: main score +16.1, code execution +25, material constraints +5.2
- Claude Sonnet 4.6: main score +15.4, code execution +24.2
- GPT-o3: main score -13.8, code execution -21.8
- Qwen3 Max: main score +12.1, code execution +32.2, material constraints -12.5
Signals to Watch
- GPT-o3: main score plummeted -13.8 points
- Gemini 3.1 Pro: material constraints plummeted -26.6 points
- GLM-4.6: incomplete data (missing communication dimension, API failure/timeout), entered automatic re-run, not ranked in this round
When reading such Smoke briefs, the focus should be on two questions: first, whether a model has been exposed to the same type of weakness for multiple consecutive days; second, whether the integrity rating has changed from pass to warn or fail. Large daily changes in execution or constraint scores may stem from question sampling or be early signals of real degradation, requiring subsequent runs to verify.
Data source: YZ Index | Run #237 | View raw data
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接