The 2026-07-08 YZ Index Smoke Quick Test covered 10 models, with DeepSeek V4 Pro ranking first at 95.19 points that day. Smoke is a daily 10-question quick test suitable for observing short-term signals and is not equivalent to the Full weekly ranking conclusions.
This Smoke evaluation only covers two main leaderboard dimensions: Code Execution and Material Constraints. The main leaderboard 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 making long-term conclusions about model capabilities.
Daily Ranking
| Rank | Model | Main Leaderboard | Code Execution | Material Constraints | Integrity |
|---|---|---|---|---|---|
| #1 | DeepSeek V4 Pro | 95.19 | 100 | 89.3 | pass |
| #2 | Grok 4 | 91.69 | 98.3 | 83.6 | pass |
| #3 | Claude Sonnet 4.6 | 86.28 | 100 | 69.5 | pass |
| #4 | GPT-o3 | 83.44 | 83.3 | 83.6 | pass |
| #5 | Claude Opus 4.7 | 81.44 | 75 | 89.3 | pass |
| #6 | Qwen3 Max | 77.83 | 73.1 | 83.6 | pass |
| #7 | 豆包 Pro | 71.83 | 83.3 | 57.8 | pass |
| #8 | GPT-5.5 | 66.09 | 75 | 55.2 | pass |
| #9 | Gemini 3.1 Pro | 61 | 58.3 | 64.3 | pass |
| #10 | Gemini 2.5 Pro | 59.9 | 56.3 | 64.3 | pass |
Data Interpretation
In today's YZ Index Smoke Quick Test, DeepSeek V4 Pro achieved a main leaderboard score of 95.19 with a combination of 100 in Code Execution and 89.3 in Material Constraints, ranking first; Grok 4 followed closely with a main leaderboard score of 91.69, thanks to 98.3 in Code Execution and 83.6 in Material Constraints; Claude Sonnet 4.6 secured 86.28 on the main leaderboard with 100 in Code Execution and 69.5 in Material Constraints. All three exhibit a structural characteristic where Code Execution is stronger than Material Constraints. Among them, DeepSeek V4 Pro's Material Constraints score of 89.3 and Grok 4's 83.6 complement each other, supporting the top two positions on the main leaderboard.
GPT-5.5 dropped 16.7 points on the main leaderboard, with a 25-point decline in Code Execution and a 6.5-point decline in Material Constraints; Claude Opus 4.7 dropped 15.6 points on the main leaderboard, with a 25-point decline in Code Execution; Gemini 2.5 Pro dropped 9 points on the main leaderboard, with a 42.2-point decline in Code Execution and a 31.5-point increase in Material Constraints. Qwen3 Max rose 16.5 points on the main leaderboard, with a 40.3-point increase in Material Constraints; DeepSeek V4 Pro rose 15 points on the main leaderboard, with a 31.5-point increase in Material Constraints. These single-day fluctuations may stem from question sampling variability or could reflect actual performance differences of the models under specific constraints, requiring subsequent same-methodology run re-checks for confirmation.
Overall, under the main leaderboard formula, the strength combination of Code Execution and Material Constraints directly affects the ranking. As a small-sample single-day signal, the Smoke Quick Test only reflects the day's data distribution and does not provide long-term trend judgment.
Major Changes
- GPT-5.5: Main Leaderboard -16.7, Code Execution -25, Material Constraints -6.5
- Qwen3 Max: Main Leaderboard +16.5, Material Constraints +40.3
- Claude Opus 4.7: Main Leaderboard -15.6, Code Execution -25
- DeepSeek V4 Pro: Main Leaderboard +15, Material Constraints +31.5
- Gemini 2.5 Pro: Main Leaderboard -9, Code Execution -42.2, Material Constraints +31.5
Signals to Watch
- No publishable anomaly signals were retained in this test.
When reading this type of Smoke brief, the focus should be on two issues: first, whether a model exposes the same type of weakness on consecutive days; second, whether the integrity rating drops from pass to warn or fail. Significant daily fluctuations in execution or constraint scores may stem from question sampling or could be early signals of actual degradation, requiring subsequent run re-checks.
Data Source: YZ Index | Run #220 | View Raw Data
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接