Claude Opus 4.7, Claude Sonnet 4.6, DeepSeek V4 Pro, and Grok 4 Tied at 96.99: 2026-07-12 Smoke Quick Test Data Brief

On 2026-07-12, the YZ Index Smoke Quick Test covered 11 models. Claude Opus 4.7, Claude Sonnet 4.6, DeepSeek V4 Pro, and Grok 4 tied for first place with a score of 96.99. Smoke is a daily 10-question quick test suitable for observing short-term signals, not equivalent to the Full weekly ranking conclusions.

This Smoke evaluation only covers two main ranking dimensions: Code Execution and Material Constraint. The main ranking 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 Ranking

Data Interpretation

In today's YZ Index Smoke Quick Test, four models—Claude Opus 4.7, Claude Sonnet 4.6, DeepSeek V4 Pro, and Grok 4—shared the same main rank of 96.99, with Code Execution at 100 and Material Constraint at 93.3, demonstrating a balanced high-level combination of both dimensions. Gemini 2.5 Pro and Qwen3 Max both had a main rank of 94.24, with Code Execution at 95 and Material Constraint at 93.3, also maintaining strong code execution performance under material constraints.

Compared to the previous run under the same methodology, Gemini 2.5 Pro's main rank rose by 33.4 points and Code Execution by 57.5 points; DeepSeek V4 Pro's main rank rose by 29.3 points and Code Execution by 50 points; Grok 4's main rank rose by 22.4 points and Code Execution by 37.5 points; Qwen3 Max's main rank rose by 19.7 points and Code Execution by 32.5 points. GLM-4.6, however, saw a sharp drop of -25 points in Material Constraint, with a main rank of only 52.5 and Material Constraint at 25. Such a one-day anomaly could stem from question sampling fluctuations or reflect genuine degradation, requiring confirmation in subsequent runs.

The Smoke Quick Test is a small-sample single-day signal. The above score structure and changes are for daily observation only and do not constitute long-term conclusions.

Key Changes

• Gemini 2.5 Pro: Main rank +33.4, Code Execution +57.5
• DeepSeek V4 Pro: Main rank +29.3, Code Execution +50
• Grok 4: Main rank +22.4, Code Execution +37.5
• Qwen3 Max: Main rank +19.7, Code Execution +32.5
• GLM-4.6: Main rank +16.3, Code Execution +50, Material Constraint -25

Signal to Watch

• GLM-4.6: Material Constraint dropped sharply by -25 points

When reading Smoke briefs like this, the focus should be on two questions: first, whether a model continuously exposes the same weakness over multiple days; second, whether the integrity rating shifts from pass to warn or fail. Large daily swings in execution or constraint scores may come from question sampling or could be early signs of genuine degradation, requiring confirmation in subsequent runs.

Data Source: YZ Index | Run #226 | View Raw Data