GPT-6 Astra and GPT-6 Sol Tie at 84.9 Points: 2026-11-02 Smoke Quick Test Data Brief

The 2026-11-02 Winzheng Index Smoke quick test covered 14 models, with GPT-6 Astra and GPT-6 Sol tying for first place of the day at 84.9 points. Smoke is a daily 10-question quick test, suited to observing short-term signals and not equivalent to the conclusions of the weekly Full ranking.

This Smoke evaluation covered only two main leaderboard dimensions, code execution and material constraints, with the main leaderboard formula being 0.55 × code execution + 0.45 × material constraints. Because the daily sample size is small, single-day scores are better treated as monitoring signals rather than long-term conclusions about model capability.

Daily Rankings

Data Interpretation

Today's top two on the main leaderboard, GPT-6 Astra and GPT-6 Sol, both recorded 84.9 on the main leaderboard, with code execution at 75 and material constraints at 97, indicating a balanced combination of the two capabilities. Claude Opus 4.7 scored 82.84 on the main leaderboard, with code execution at 68.8 and material constraints at 100; its material constraints score was higher than the top two, but its code execution was lower. Gemini 2.5 Pro scored 74.29 on the main leaderboard, with code execution at 73.7 and material constraints at 75, putting code execution close to the leaders while material constraints were relatively low.

Compared with the previous run under the same methodology, Gemini 2.5 Pro rose 24.3 points on the main leaderboard, with code execution up 23.7 points and material constraints up 25 points; GLM-4.6 fell 25 points on the main leaderboard, with both code execution and material constraints down 25 points; GPT-o3 fell 18.5 points on the main leaderboard, with code execution down 39.1 points and material constraints up 6.7 points. These changes may stem from question sampling fluctuations or may reflect genuine performance degradation, and need to be confirmed by follow-up runs.

DeepSeek V4 Pro and Claude Opus 4.7 likewise showed double-digit declines on the main leaderboard, with code execution dropping by 50 points and 29 points respectively, while material constraints moved in different directions. As a small-sample single-day signal, the structural features above from Smoke are for reference in observing the day only.

Main Changes

• GLM-4.6: main leaderboard down 25 points, code execution −25 points, material constraints −25 points
• Gemini 2.5 Pro: main leaderboard up 24.3 points, code execution +23.7 points, material constraints +25 points
• GPT-o3: main leaderboard down 18.5 points, code execution −39.1 points, material constraints +6.7 points
• DeepSeek V4 Pro: main leaderboard down 16.3 points, code execution −50 points, material constraints +24.9 points
• Claude Opus 4.7: main leaderboard down 16 points, code execution −29 points

Signals to Watch

• No publishable anomaly signals were retained in this run.

When reading this kind of Smoke brief, the focus should be on two questions: first, whether a given model exposes the same type of weakness on consecutive days; second, whether its integrity rating moves from pass into warn or fail. Large single-day swings in execution or constraint scores may come from question sampling, or may be an early signal of genuine degradation, and require follow-up runs for confirmation.

Data source: YZ Index | Run #407 | View raw data