SimpleBench: leaderboard
Evaluates 'linguistic adversarial robustness' - ability to answer trick questions. 200+ questions where everyday human reasoning (83.7%) still outperforms frontier LLMs on spatio-temporal reasoning and social intelligence.
Metric: Score (AVG@5). Source: simple-bench.com. Saturation forecast: Around January 2027. 106 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 | 88.4 | #3 |
| 2 | Claude Fable 5.1 | 86.6 | #9 |
| 3 | Median Human | 83.7 | |
| 4 | GPT-6 Astra | 83.6 | #6 |
| 5 | GPT-6.1 Sol | 82.9 | #7 |
| 6 | Gemini 3.8 Flash | 82.4 | #21 |
| 7 | Claude Fable 5 | 81.9 | #16 |
| 8 | Muse Spark 1.3 | 81.8 | #24 |
| 9 | Claude Opus 5 | 80.6 | #12 |
| 10 | Gemini 3.1 Pro (Preview) | 79.6 | #58 |
| 11 | GPT-5.5 Pro | 76.9 | #17 |
| 12 | Gemini 3.5 Flash | 76.7 | #60 |
| 13 | Gemini 3 Pro (Preview) | 76.4 | #66 |
| 14 | Grok 4.6 | 75.9 | #26 |
| 15 | Claude Sonnet 5.5 | 75.9 | #11 |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=simplebench · How It Works · Data refreshed daily, snapshot 2026-10-20.