Gemini 2.5 Pro and Gemini 3.1 Pro Tie at 92.44: 2026-07-17 Smoke Quick Test Data Brief

On July 17, 2026, the YZ Index Smoke quick test covered 11 models, with Gemini 2.5 Pro and Gemini 3.1 Pro tying for first place at 92.44 points. Smoke is a daily 10-question quick test suitable for observing short-term signals and is not equivalent to Full weekly rankings.

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

Daily Rankings

Data Interpretation

In today's YZ Index Smoke quick test, Gemini 2.5 Pro and Gemini 3.1 Pro tied for first place on the main board with 92.44 points, Code Execution at 95 and Material Constraints at 89.3, showing a clear advantage from their strong combination in both dimensions. GLM-4.6 achieved a main board score of 84.68, driven by Code Execution at 91.7, while Claude Opus 4.7's main board score of 84.52 relied primarily on Material Constraints at 89.3, creating complementary structures at the top. GPT-o3 scored 89.5 on Code Execution but only 63.6 on Material Constraints, yielding an overall score of 77.85, reflecting how the Material Constraints weakness limits its overall ranking.

Compared to the previous same-scope run, Gemini 2.5 Pro's main board rose by 33.9 points and Code Execution by 61.7 points; GLM-4.6's main board rose by 13.5 points and Code Execution by 33.4 points; Gemini 3.1 Pro's main board rose by 12 points and Code Execution by 20 points. The improvement in Code Execution scores for these models directly drove the main board changes. Grok 4's main board fell by 17.5 points, Code Execution by 13.9 points, and Material Constraints by 21.9 points; DeepSeek V4 Pro's main board fell by 11.9 points, Code Execution by 13.3 points, and Material Constraints by 10.2 points; Claude Sonnet 4.6's main board fell by 8.4 points. These movements occurred within a single day's small sample data.

The sharp main board drops observed in DeepSeek V4 Pro, Grok 4, and Claude Sonnet 4.6 may stem from question sampling fluctuations or represent real performance degradation under specific constraints, requiring subsequent runs to confirm signal stability. As a small-sample single-day signal, Smoke quick test interpretations are limited to today's data comparison and should not be used for long-term inferences.

Key Changes

• Gemini 2.5 Pro: Main board +33.9, Code Execution +61.7
• Grok 4: Main board -17.5, Code Execution -13.9, Material Constraints -21.9
• GLM-4.6: Main board +13.5, Code Execution +33.4, Material Constraints -10.9
• Gemini 3.1 Pro: Main board +12, Code Execution +20
• DeepSeek V4 Pro: Main board -11.9, Code Execution -13.3, Material Constraints -10.2

Signals to Watch

• DeepSeek V4 Pro: Main board sharp drop of -11.9 points
• Grok 4: Main board sharp drop of -17.5 points
• Claude Sonnet 4.6: Main board sharp drop of -8.4 points

When reading Smoke briefs like this, focus on two questions: first, whether a model consistently exposes the same type of weakness over multiple consecutive days; second, whether the Integrity rating shifts from pass to warn or fail. Large single-day changes in Execution or Constraints scores may result from question sampling or be early signals of real degradation, requiring subsequent runs to confirm.

Data source: YZ Index | Run #235 | View Raw Data