LLM2014 Logic 2026-08: leaderboard
An August 2026 archive of a personal Chinese-language project that keeps a private, monthly rotating bank of at most 28 logic problems and 270 test cases. Each problem scores out of ten on its scoring points, with nothing for a guessed answer or a reply that breaks the stated format. Every problem is answered three times and the table reports maximum and median columns; this board takes the median, higher is better.
Metric: Median Score. Source: raw.githubusercontent.com. Saturation forecast: Around October 2027. 46 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (xHigh) | 73.89 | #31 (GPT-5.5) |
| 2 | GPT-5.6 Sol (xHigh) | 69.42 | #19 (GPT-5.6 Sol) |
| 3 | Kimi K3 (Max) | 67.66 | #30 (Kimi K3) |
| 4 | Claude Opus 5 (xHigh) | 64.74 | #12 (Claude Opus 5) |
| 5 | GLM-5.3 (Max) | 63.91 | #39 (GLM-5.3) |
| 6 | GLM-5.3 Flash (Max) | 60.52 | #65 (GLM-5.3 Flash) |
| 7 | Qwen 3.8 Max (xHigh) | 60.05 | #35 (Qwen 3.8 Max) |
| 8 | DeepSeek V4 Pro (0813) (Max) | 59.63 | #53 (DeepSeek V4 Pro (0813)) |
| 9 | DeepSeek-V4-Flash-Vision-Exp (Max) | 58.1 | #76 (DeepSeek-V4-Flash-Vision-Exp) |
| 10 | Gemini 3.7 Flash (High) | 57.56 | #23 (Gemini 3.7 Flash) |
| 11 | Grok 4.6 (High) | 56.41 | #26 (Grok 4.6) |
| 12 | Gemini 3.1 Pro (Preview) (High) | 55.84 | #58 (Gemini 3.1 Pro (Preview)) |
| 13 | DeepSeek V4 Flash (0731) (Max) | 55.23 | #78 (DeepSeek V4 Flash (0731)) |
| 14 | GLM-5.2 (Max) | 54.7 | #73 (GLM-5.2) |
| 15 | GPT-5.6 Luna (xHigh) | 51.62 | #68 (GPT-5.6 Luna) |
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-2026-08 · How It Works · Data refreshed daily, snapshot 2026-10-20.