LLM2014 Logic 2026-07: leaderboard
July 2026 archive of the logic set from a personal Chinese tracker whose private question bank rotates monthly and stays within 28 problems and 270 cases. Points are given per scoring point out of ten per problem, and answers that break the requested output format or reach the answer without correct reasoning score zero; each problem is run three times. The board takes the median score column, higher is better.
Metric: Median Score. Source: raw.githubusercontent.com. Saturation forecast: Around January 2027. 45 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (xHigh) | 77.46 | #31 (GPT-5.5) |
| 2 | Kimi K3 (Max) | 74.8 | #30 (Kimi K3) |
| 3 | GPT-5.6 Sol (xHigh) | 72.99 | #19 (GPT-5.6 Sol) |
| 4 | Claude Opus 5 (xHigh) | 71.88 | #12 (Claude Opus 5) |
| 5 | DeepSeek V4 Flash (0731) (Max) | 58.8 | #78 (DeepSeek V4 Flash (0731)) |
| 6 | Gemini 3.1 Pro (Preview) (High) | 58.46 | #58 (Gemini 3.1 Pro (Preview)) |
| 7 | GLM-5.2 (Max) | 58.27 | #73 (GLM-5.2) |
| 8 | Muse Spark 1.1 | 57.23 | #52 |
| 9 | Grok 4.5 (High) | 56.72 | #48 (Grok 4.5) |
| 10 | GPT-5.6 Luna (xHigh) | 51.97 | #68 (GPT-5.6 Luna) |
| 11 | Hy3 (High) | 44.6 | #93 (Hy3) |
| 12 | Claude Sonnet 5 (xHigh) | 43.14 | #57 (Claude Sonnet 5) |
| 13 | Gemini 3.6 Flash (High) | 41.74 | #51 (Gemini 3.6 Flash) |
| 14 | MiniMax-M3 | 40.94 | #136 |
| 15 | Claude Opus 5 | 31.73 | #12 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm2014-logic-2026-07 · How It Works · Data refreshed daily, snapshot 2026-10-20.