GPT-5.5 Leads With 96.1 Points: 2026-10-22 Smoke Quick-Test Data Brief

The 2026-10-22 YZ Index Smoke quick test covered 14 models, with GPT-5.5 ranking first for the day at 96.1 points. Smoke is a daily 10-question quick test, suitable for observing short-term signals; it is not equivalent to a Full weekly leaderboard conclusion.

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

Rankings for the Day

Reading the Data

In today's YZ Index Smoke quick test, GPT-5.5 ranked first with a main leaderboard score of 96.1, pairing 97 in code execution with 95 in material constraint for a balanced profile; Doubao Pro scored 93.09 on the main leaderboard, also reaching 97 in code execution but 88.3 in material constraint; Gemini 2.5 Pro scored 92.49 on the main leaderboard, with 100 in code execution and 83.3 in material constraint. A cluster of models that followed — Gemini 3.1 Pro, GPT-6 Luna, GPT-6.1 Sol, GPT-o3 and Grok 4 — all landed at 90.84 on the main leaderboard, with code execution concentrated at 97 and material constraint at 83.3, showing a structure dominated by code execution. Claude Opus 4.7 scored 82.35 on the main leaderboard, displaying an inverse pairing of 72 in code execution and 95 in material constraint.

Compared with the previous run on the same basis, GLM-4.6 showed a main leaderboard change of -57.5 points, with code execution at -45 and material constraint at -72.8; GPT-6.1 Sol rose 28.5 points on the main leaderboard and 52.2 in code execution; GPT-o3 rose 24.6 points on the main leaderboard and 47 in code execution; GPT-5.5 rose 17.6 points on the main leaderboard, 27 in code execution and 6.1 in material constraint; GPT-6 Luna rose 17.5 points on the main leaderboard and 32.2 in code execution. These swings may stem from question sampling variance, or they may reflect genuine degradation in model performance, and require follow-up runs to confirm. Gemini 3.1 Pro's material constraint plunge of -16.7 points and Claude Sonnet 4.6's main leaderboard plunge of -16.9 points likewise point to potential anomalies, while GLM-4.6's main leaderboard plunge of -57.5 points also awaits further verification.

The Smoke quick test is a small-sample, single-day signal. Qwen3 Max did not participate in this period's rankings because an API failure left its data incomplete. Overall, the observations show differences among leading models in how code execution and material constraint pair up; anomalous score changes should be interpreted cautiously to avoid over-extrapolation.

Major Changes

• GLM-4.6: main leaderboard down 57.5 points, code execution -45, material constraint -72.8
• GPT-6.1 Sol: main leaderboard up 28.5 points, code execution +52.2
• GPT-o3: main leaderboard up 24.6 points, code execution +47
• GPT-5.5: main leaderboard up 17.6 points, code execution +27, material constraint +6.1
• GPT-6 Luna: main leaderboard up 17.5 points, code execution +32.2

Signals to Watch

• Gemini 3.1 Pro: material constraint plunges -16.7 points
• Claude Sonnet 4.6: main leaderboard plunges -16.9 points
• GLM-4.6: main leaderboard plunges -57.5 points
• Qwen3 Max: incomplete data (several evaluation dimensions missing due to API failure/timeout); automatically queued for a rerun, and not included in this period's rankings

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 across several consecutive days; second, whether its integrity rating moves from pass to warn or fail. Large single-day swings in execution or constraint scores may come from question sampling, or they may be an early signal of genuine degradation, and require follow-up runs to confirm.

Data source: YZ Index | Run #391 | View raw data