Claude Opus 4.7 Tops the List with 84.9 Points: 2026-10-21 Smoke Quick-Test Data Brief

The 2026-10-21 YZ Index Smoke quick test covered 14 models, with Claude Opus 4.7 taking first place for the day at 84.9 points. Smoke is a daily 10-question quick test suited to spotting short-term signals, and it is not equivalent to the conclusions of the weekly Full leaderboard.

This Smoke evaluation covers only two Main Leaderboard dimensions: Code Execution and Material Constraints. The Main Leaderboard formula is 0.55 × Code Execution + 0.45 × Material Constraints. Because the daily sample is small, single-day scores are better treated as monitoring signals than as long-term conclusions about model capability.

Daily Rankings

Data Interpretation

Today's Smoke leaders show a pattern of "high Material Constraints, adequate Code Execution." Claude Opus 4.7 has a Main Leaderboard score of 84.9, with Code Execution at 75 and Material Constraints at 97, a relatively balanced advantage; Gemini 3.1 Pro has a Main Leaderboard score of 83.5, with Code Execution at 70 and Material Constraints at 100, relying more on the material side for its lift; GLM-4.6 has a Main Leaderboard score of 82.51, with Code Execution at 70 and Material Constraints at 97.8, a similar structure.

The upper-middle tier shows a slight divergence in emphasis: Claude Sonnet 4.6, DeepSeek V4 Pro, and Grok 4 all share a Main Leaderboard score of 80.99, with Code Execution at 75 and Material Constraints at 88.3, making them more solid on the code side; Doubao Pro has a Main Leaderboard score of 79.65 and GPT-6 Sol 79.29, both with Code Execution at 64.8, but with Material Constraints at 97.8 and 97 respectively, indicating strong adherence on materials while their Main Leaderboard scores are held back by Code Execution.

In terms of notable moves, GPT-o3 fell 31.6 points on the Main Leaderboard, 48.5 points on Code Execution, and 10.9 points on Material Constraints; GPT-6.1 Sol fell 22.5 points on the Main Leaderboard and 52.2 points on Code Execution while gaining 13.9 points on Material Constraints, with the pressure concentrated on Code Execution; DeepSeek V4 Pro rose 21.4 points on the Main Leaderboard, 28 points on Code Execution, and 13.3 points on Material Constraints, looking more like a recovery on both fronts. Anomaly signals: (none). The changes may reflect sampling variation in the questions, or genuine degradation or recovery; Smoke is a single-day small sample, so follow-up runs are needed for confirmation.

Key Changes

• GPT-o3: Main Leaderboard down 31.6 points, Code Execution -48.5 points, Material Constraints -10.9 points
• GPT-6.1 Sol: Main Leaderboard down 22.5 points, Code Execution -52.2 points, Material Constraints +13.9 points
• DeepSeek V4 Pro: Main Leaderboard up 21.4 points, Code Execution +28 points, Material Constraints +13.3 points
• Gemini 3.1 Pro: Main Leaderboard up 19.8 points, Code Execution +23 points, Material Constraints +15.9 points
• GLM-4.6: Main Leaderboard up 18.8 points, Code Execution +23 points, Material Constraints +13.7 points

Signals to Watch

• No publishable anomaly signals were retained for this run.

When reading this kind of Smoke brief, the focus should be on two questions: first, whether a model repeatedly shows the same type of weakness over several consecutive days; second, whether its Integrity rating moves from pass to warn or fail. Large single-day changes in execution or constraint scores may come from question sampling, or they may be early signals of genuine degradation, and follow-up runs are needed for confirmation.

Data source: YZ Index | Run #389 | View raw data