GPT-6 Astra and GPT-6.1 Sol Tie at 81.66: 2026-11-05 Smoke Quick Test Data Brief

The 2026-11-05 YZ Index Smoke quick test covered 13 models, with GPT-6 Astra and GPT-6.1 Sol tied for first place that day at 81.66 points. Smoke is a daily 10-question quick test, suitable for observing short-term signals, and is not equivalent to Full weekly leaderboard conclusions.

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 size is small, a single-day score is better used as a monitoring signal rather than a long-term conclusion about model capability.

Daily Ranking

Data Interpretation

In today's YZ Index Smoke quick test, GPT-6 Astra and GPT-6.1 Sol tied at 81.66 points on the main leaderboard; both scored 75 in code execution and 89.8 in material constraints, showing a balanced pairing of code execution and material constraints. DeepSeek V4 Pro and Gemini 3.1 Pro followed closely, both with 77.1 points on the main leaderboard; their structure of 66.7 in code execution and 89.8 in material constraints also highlights the advantage in material constraints. GPT-6 Sol's 75.13 main leaderboard score comes from a relatively balanced combination of 74.5 in code execution and 75.9 in material constraints, while Doubao Pro's 70.16 main leaderboard score relies on a material-constraint-dominated pattern of 50 in code execution and 94.8 in material constraints.

Claude Opus 4.7 dropped 42.4 points on the main leaderboard, 68.7 points in code execution, and 10.2 points in material constraints; GPT-5.5 dropped 40.6 points on the main leaderboard and 74.3 points in code execution; GPT-o3 dropped 38.4 points on the main leaderboard and 71.1 points in code execution. These significant changes may stem from question sampling fluctuations or may reflect real single-day performance degradation. Gemini 3.1 Pro rose 36.5 points on the main leaderboard, 35.9 points in code execution, and 37.2 points in material constraints, while DeepSeek V4 Pro rose 34.4 points on the main leaderboard, 41.7 points in code execution, and 25.5 points in material constraints; these also require subsequent same-scope run review to confirm signal stability.

The Smoke quick test is a small-sample, single-day signal. Top models in the 89.8-point material constraints range maintain relatively high main leaderboard performance, while differences in the strong/weak pairing of code execution and material constraints directly affect the core_overall calculation. All interpretations are based on today's data, avoiding conclusions about models' long-term status.

Key Changes

• Claude Opus 4.7: main leaderboard down 42.4 points, code execution -68.7 points, material constraints -10.2 points
• GPT-5.5: main leaderboard down 40.6 points, code execution -74.3 points
• GPT-o3: main leaderboard down 38.4 points, code execution -71.1 points
• Gemini 3.1 Pro: main leaderboard up 36.5 points, code execution +35.9 points, material constraints +37.2 points
• DeepSeek V4 Pro: main leaderboard up 34.4 points, code execution +41.7 points, material constraints +25.5 points

Signals to Watch

• No publishable anomalous signals were retained in this run.

When reading this type of Smoke brief, the focus should be on two questions: first, whether a model has exposed the same type of weakness for multiple consecutive days; second, whether its integrity rating has moved from pass to warn or fail. Large single-day changes in execution or constraint scores may come from question sampling or may be early signals of real degradation and require subsequent run review.

Data source: YZ Index | Run #413 | View raw data