LLM2014 Logic 2025-12: leaderboard
LLM2014 monthly logic benchmark for 2025-12, reporting aggregate logical-reasoning performance on the source score scale.
Metric: Median Score. Source: raw.githubusercontent.com. Saturation forecast: Around January 2027. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (High) | 81.83 | #110 (GPT-5.2) |
| 2 | Gemini 3 Pro | 79.29 | #80 |
| 3 | GPT-5.1 (High) | 77.54 | #135 (GPT-5.1) |
| 4 | Gemini 3 Flash (High) | 76.9 | #99 (Gemini 3 Flash) |
| 5 | GPT-5 Mini (High) | 75.05 | #186 (GPT-5 Mini) |
| 6 | DeepSeek V3.2 Speciale | 72.68 | #128 |
| 7 | Grok 4 | 65.61 | #181 |
| 8 | GLM-4.7 (Thinking) | 65.49 | #192 (GLM-4.7) |
| 9 | Claude Sonnet 4.5 (Thinking) | 62.7 | #142 (Claude Sonnet 4.5) |
| 10 | Kimi K2 (Thinking) | 59.34 | #243 (Kimi K2) |
| 11 | Grok 4.1 Fast (Reasoning) | 57.68 | #217 (Grok 4.1 Fast) |
| 12 | Qwen 3 Max (Thinking) | 49.68 | #208 (Qwen 3 Max) |
| 13 | Qwen 3 235B A22B 2507 (Thinking) | 46.71 | #260 (Qwen 3 235B A22B 2507) |
| 14 | Qwen 3 Max | 42.85 | #208 |
| 15 | GPT-OSS-120B (High) | 41.96 | #341 (GPT-OSS-120B) |
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-2025-12 · How It Works · Data refreshed daily, snapshot 2026-10-20.