A2RBench: leaderboard
Abstract reasoning ability reflects the intelligence and generalization capacity of LLMs to extract and apply abstract rules. However, accurately measuring this ability remains challenging: existing benchmarks either rely on expensive manual annotation, limiting their scale, or risk measuring memorization rather than genuine reasoning. To address this, we introduce an automated pipeline named A2RB
Metric: Total Acc (self-reported). Source: benchmarklist.com. Saturation forecast: Around April 2027. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 40.9 | #80 |
| 2 | GPT-5 | 39 | #98 |
| 3 | GPT-5 Mini | 36.9 | #186 |
| 4 | O4 Mini | 33.7 | #180 |
| 5 | Claude Sonnet 4.5 | 30.6 | #142 |
| 6 | GPT-5.2 | 28.3 | #110 |
| 7 | Gemini 2.5 Flash | 27.7 | #245 |
| 8 | GLM-4.6 | 21 | #252 |
| 9 | DeepSeek V3.2 | 20.8 | #205 |
| 10 | Qwen 3 32B | 19.3 | #443 |
| 11 | GPT-4o Mini | 16.1 | #599 |
| 12 | Qwen 3 14B | 14.2 | #537 |
| 13 | Gemini 3 Flash | 13.1 | #99 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=a2rbench · How It Works · Data refreshed daily, snapshot 2026-10-20.