LLM2014 Logic 2025-09: leaderboard
Metric: Median Score. Source: raw.githubusercontent.com. 46 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 (High) | 84.1 |
| 2 | GPT-5 Mini (High) | 78.12 |
| 3 | Grok 4 | 74.79 |
| 4 | Grok 4 Fast (Reasoning) | 67.59 |
| 5 | DeepSeek V3.1 (Thinking) | 54.13 |
| 6 | GPT-OSS-120B (High) | 53.51 |
| 7 | Qwen 3 235B A22B 2507 (Thinking) | 53.11 |
| 8 | Gemini 2.5 Pro | 51.65 |
| 9 | Claude Sonnet 4 (Thinking) | 48.04 |
| 10 | Qwen 3 30B A3B 2507 (Thinking) | 46.52 |
| 11 | Qwen 3 Next 80B A3B (Thinking) | 43.59 |
| 12 | Qwen 3 235B A22B 2507 | 43.08 |
| 13 | Qwen 3 Max (Preview) | 42.33 |
| 14 | Claude Opus 4 | 40.23 |
| 15 | Kimi K2 0905 | 40.01 |
Interactive version: theaggregate.ai/benchmark?slug=llm2014-logic-2025-09 · How It Works · Data refreshed daily, snapshot 2026-09-05.