LLM2014 Logic 2025-10: leaderboard
Metric: Median Score. Source: raw.githubusercontent.com. 50 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 (High) | 86.25 |
| 2 | GPT-5 Mini (High) | 79.37 |
| 3 | Grok 4 | 74.97 |
| 4 | Claude Sonnet 4.5 (Thinking) | 69.78 |
| 5 | Grok 4 Fast (Reasoning) | 67.76 |
| 6 | Qwen 3 235B A22B 2507 (Thinking) | 54.18 |
| 7 | DeepSeek V3.2 (Thinking) | 54.16 |
| 8 | GPT-OSS-120B (High) | 53.24 |
| 9 | Qwen 3 Max | 52.61 |
| 10 | Gemini 2.5 Pro | 52.54 |
| 11 | Qwen 3 30B A3B 2507 (Thinking) | 47.41 |
| 12 | Ring-1T | 46.07 |
| 13 | Qwen 3 235B A22B 2507 | 43.97 |
| 14 | Qwen 3 Next 80B A3B (Thinking) | 43.77 |
| 15 | Claude Opus 4 | 40.23 |
Interactive version: theaggregate.ai/benchmark?slug=llm2014-logic-2025-10 · How It Works · Data refreshed daily, snapshot 2026-09-05.