Claude Opus 4.7 Leads with 81.91 Points: 2026-10-18 Smoke Quick Test Data Brief

The 2026-10-18 YZ Index Smoke quick test covered 14 models, with Claude Opus 4.7 taking the top spot for the day at 81.91 points. Smoke is a daily 10-question quick test, suited to observing short-term signals and not equivalent to the conclusions of the Full weekly leaderboard.

This Smoke evaluation covered 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 than as a long-term verdict on model capability.

Rankings for the Day

Reading the Data

Looking at today's YZ Index Smoke quick test data, the leading models show distinct profiles in how code execution and material constraints pair up. Claude Opus 4.7 scores 81.91 on the main leaderboard, with 83.3 in code execution and 80.2 in material constraints, a fairly balanced pairing; Doubao Pro scores 77.34 on the main leaderboard, with 75 in code execution and 80.2 in material constraints, relatively stronger on material constraints; Grok 4 scores 72.09 on the main leaderboard, with 72.9 in code execution and 71.1 in material constraints, a similar structure; Gemini 2.5 Pro scores 70.45 on the main leaderboard, with 78.1 in code execution and 61.1 in material constraints, showing a clear advantage in code execution but a comparatively low material constraints score; DeepSeek V4 Pro and GPT-5.5 both score 68.16 on the main leaderboard, each with 58.3 in code execution and 80.2 in material constraints, standing out on material constraints.

Compared with the previous run under the same methodology, GPT-o3 fell 39.2 points on the main leaderboard, with code execution down 55.2 and material constraints down 19.7; GPT-6.1 Sol fell 28.3 points on the main leaderboard, with code execution down 30.2 and material constraints down 25.9; Gemini 3.1 Pro fell 23.7 points on the main leaderboard, with code execution down 41.7; GPT-6 Luna fell 21.4 points on the main leaderboard, with code execution down 17.7 and material constraints down 25.9; GPT-6 Astra fell 19.6 points on the main leaderboard, with code execution down 25 and material constraints down 12.9. These changes may stem from sampling variation in the questions, or may reflect genuine degradation in model performance, and require follow-up runs to confirm.

As a small-sample single-day signal, the observations above reflect only the data structure of the day; no long-term judgment is made at this time.

Key Changes

• GPT-o3: main leaderboard down 39.2 points, code execution −55.2, material constraints −19.7
• GPT-6.1 Sol: main leaderboard down 28.3 points, code execution −30.2, material constraints −25.9
• Gemini 3.1 Pro: main leaderboard down 23.7 points, code execution −41.7
• GPT-6 Luna: main leaderboard down 21.4 points, code execution −17.7, material constraints −25.9
• GPT-6 Astra: main leaderboard down 19.6 points, code execution −25, material constraints −12.9

Signals to Watch

• No publishable anomaly signals were retained this run.

When reading this kind of Smoke brief, the focus should be on two questions: first, whether a given model shows the same type of weakness across 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 may be an early signal of genuine degradation, and require follow-up runs to confirm.

Data source: YZ Index | Run #383 | View raw data