GPT-o3 Tops with 86.9 Points: 2026-07-10 Smoke Quick Test Data Brief

On 2026-07-10, the YZ Index Smoke Quick Test covered 9 models, with GPT-o3 scoring 86.9 points to take first place that day. Smoke is a daily 10-question quick test, suitable for observing short-term signals and not equivalent to Full weekly ranking conclusions.

This Smoke evaluation covers only the two main board dimensions: Code Execution and Material Constraint. The main board formula is 0.55 × Code Execution + 0.45 × Material Constraint. Due to the small daily sample size, single-day scores are better used as monitoring signals rather than long-term conclusions about model capabilities.

Daily Rankings

Data Interpretation

In today's YZ Index Smoke quick test, GPT-o3 ranked first with a main board score of 86.9, pairing a strong Code Execution of 94.5 with Material Constraint of 77.6, with the 0.55 weight on Code Execution pulling its overall lead; 豆包 Pro and Gemini 3.1 Pro tied at 84.57 on the main board, both having the same Code Execution of 93.7 and Material Constraint of 73.4, showing a balanced structure; DeepSeek V4 Pro scored 81.86 on the main board, with Code Execution 90.9 and Material Constraint 70.8, continuing its code-side advantage. Grok 4 and GPT-5.5 showed high Code Execution scores of 93.7 and 94.5 respectively, but their Material Constraint scores were only 61.7 and 59.1, dragging down their overall performance.

Claude Opus 4.7 dropped sharply by -19.3 points on the main board, with Code Execution -22 points and Material Constraint -15.9 points; Claude Sonnet 4.6 fell -15.5 points on the main board and -22 points on Code Execution; Qwen3 Max dropped -10.9 points on the main board and -22 points on Code Execution; Grok 4 fell -8.4 points on the main board, with Material Constraint -17.6 points and its Integrity shifting from pass to warn. These anomalies may stem from the single-day 10-question sampling fluctuation, or could be signals of genuine model degradation, requiring confirmation through subsequent runs with the same methodology. GLM-4.6 and Gemini 2.5 Pro had incomplete data due to API failures and were not included in this ranking.

The Smoke quick test is a small-sample single-day signal. The score structure and changes above only reflect the day's performance and should not be used for long-term inference.

Key Changes

• Claude Opus 4.7: Main Board down 19.3 points, Code Execution -22, Material Constraint -15.9
• Claude Sonnet 4.6: Main Board down 15.5 points, Code Execution -22, Material Constraint -7.6
• Qwen3 Max: Main Board down 10.9 points, Code Execution -22
• Grok 4: Main Board down 8.4 points, Material Constraint -17.6, Integrity pass→warn
• GPT-o3: Main Board up 6.5 points, Material Constraint +10.8

Signals to Watch

• Grok 4: Main Board sharply dropped -8.4 points
• Claude Opus 4.7: Main Board sharply dropped -19.3 points
• Qwen3 Max: Main Board sharply dropped -10.9 points
• Claude Sonnet 4.6: Main Board sharply dropped -15.5 points
• GLM-4.6: Incomplete data (missing judgment, integrity dimensions; API failure/timeout), has entered auto-retry, not included in this ranking
• Gemini 2.5 Pro: Incomplete data (missing execution, grounding, judgment, integrity, communication dimensions; API failure/timeout), has entered auto-retry, not included in this ranking

When reading such Smoke briefs, focus 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 shifted from pass to warn or fail. Large single-day changes in Execution or Constraint scores may stem from question sampling or could be early signals of genuine degradation, requiring confirmation through subsequent runs.

Data source: YZ Index | Run #224 | View raw data