GPT-5.5 Tops with 93.54 Points: 2026-10-19 Smoke Quick Test Data Brief

On 2026-10-19, the YZ Index Smoke quick test covered 14 models, with GPT-5.5 ranking first for the day at 93.54 points. Smoke is a daily 10-question quick test suitable for observing short-term signals; it is not equivalent to conclusions from 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 size is small, single-day scores are better used as monitoring signals rather than long-term conclusions about model capability.

Daily Ranking

Data Interpretation

In today's YZ Index Smoke quick test, GPT-5.5 ranked first on the main leaderboard with 93.54; its combination of 97 in code execution and 89.3 in material constraints created a balanced advantage. Grok 4 followed closely with 92.6 on the main leaderboard, also relying on a combination of 95.3 in code execution and 89.3 in material constraints. GPT-6 Astra, GPT-6 Sol, and GPT-6.1 Sol all scored 82.33 on the main leaderboard, with code execution also at 97 but material constraints at only 64.4, showing a structural feature weighted toward code execution. Claude Opus 4.7 scored 79.79 on the main leaderboard, and its pairing of 72 in code execution and 89.3 in material constraints presents a pattern in which material constraints are relatively more prominent.

GPT-6.1 Sol rose 28 points on the main leaderboard from the previous run under the same methodology, with code execution up 43.9 points; GPT-5.5 rose 25.4 points, with code execution up 38.7 points; GPT-6 Astra rose 23.8 points, with code execution up 38.7 points; GPT-6 Luna rose 22.3 points, with code execution up 43.9 points; Grok 4 rose 20.5 points, with code execution up 22.4 points and material constraints up 18.2 points. These single-day increases all come from small-sample tests and require follow-up runs to distinguish sampling fluctuation from actual change.

In terms of anomaly signals, DeepSeek V4 Pro plunged 18.6 points on the main leaderboard, GLM-4.6 material constraints plunged 30.2 points, and Qwen3 Max did not participate in the ranking because an API failure or timeout caused incomplete data. Under the small-sample single-day signals of Smoke, these fluctuations may stem from question sampling variation or temporary run differences; all require follow-up complete runs for confirmation and cannot support long-term judgments.

Key Changes

• GPT-6.1 Sol: main leaderboard up 28 points, code execution +43.9 points, material constraints +8.5 points
• GPT-5.5: main leaderboard up 25.4 points, code execution +38.7 points, material constraints +9.1 points
• GPT-6 Astra: main leaderboard up 23.8 points, code execution +38.7 points, material constraints +5.5 points
• GPT-6 Luna: main leaderboard up 22.3 points, code execution +43.9 points
• Grok 4: main leaderboard up 20.5 points, code execution +22.4 points, material constraints +18.2 points

Signals to Watch

• DeepSeek V4 Pro: main leaderboard plunged -18.6 points
• GLM-4.6: material constraints plunged -30.2 points
• Qwen3 Max: data incomplete due to API failure/timeout, automatically queued for a rerun, not included in this ranking

When reading this type of Smoke brief, focus on two questions: first, whether a model has repeatedly exposed the same type of weakness over multiple 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 follow-up runs for confirmation.

Data source: YZ Index | Run #385 | View raw data