# SimpleBench

Trick questions about everyday situations, space, time and people that ordinary adults usually get right but that can trip up AI. It tests common sense rather than book knowledge.

Category: reasoning. Type: leaderboard. Status: active.
Metric: Share of questions answered correctly, averaged over 5 runs (%); the human baseline is about 84% (higher is better).
Source: Epoch AI (Benchmarking Hub) — https://epoch.ai/data/benchmark_data.zip, results as of 2026-11-17, retrieved 2026-11-17, license: CC-BY-4.0.

| Rank | Model | Organization | Score AVG@5 (%) | Release date |
|---|---|---|---|---|
| 1 | Claude Fable 5.2 (unknown) | Anthropic | 94.9 | 2026-11-09 |
| 2 | Gemini 4 Carbon (unknown) | Google DeepMind | 92.6 | 2026-11-10 |
| 3 | Gemini 4 Argon (unknown) | Google DeepMind | 91.7 | 2026-10-13 |
| 4 | Claude Opus 5.5 (unknown) | Anthropic | 88.4 | 2026-09-22 |
| 5 | Claude Fable 5.1 (unknown) | Anthropic | 86.6 | 2026-09-01 |
| 6 | GPT-6 Astra (pro, max) | OpenAI | 86.5 | 2026-09-03 |
| 7 | GPT-6 Astra (unknown thinking) | OpenAI | 83.6 | 2026-09-03 |
| 8 | GPT-6.1 Sol (unknown thinking) | OpenAI | 82.9 | 2026-09-29 |
| 9 | Gemini 3.8 Flash (unknown) | Google DeepMind | 82.4 | 2026-09-02 |
| 10 | Claude Fable 5 (max) | Anthropic | 81.9 | 2026-06-09 |
| 11 | Muse Spark 1.3 (unknown) | Meta AI | 81.8 | 2026-09-02 |
| 12 | Claude Opus 5 (unknown) | Anthropic | 80.6 | 2026-07-24 |
| 13 | Gemini 3.1 Pro Preview | Google DeepMind | 79.6 | 2026-02-19 |
| 14 | GPT-5.5 Pro (unknown thinking) | OpenAI | 76.9 | 2026-04-23 |
| 15 | GPT-5.5 Pro (high) | OpenAI | 76.9 | 2026-04-23 |
| 16 | Gemini 3.5 Flash | Google DeepMind | 76.7 | 2026-05-19 |
| 17 | Gemini 3 Pro Preview | Google DeepMind | 76.4 | 2025-11-18 |
| 18 | Grok 4.6 (unknown) | xAI | 75.9 | 2026-08-12 |
| 19 | Claude Sonnet 5.5 (unknown) | Anthropic | 75.9 | 2026-09-28 |
| 20 | Muse Spark 1.2 | Meta AI | 74.5 | 2026-08-05 |
| 21 | GPT-5.4 Pro | OpenAI | 74.1 | 2026-03-05 |
| 22 | GPT-6 Sol (unknown thinking) | OpenAI | 73.1 | 2026-09-22 |
| 23 | Muse Ember (unknown) | Meta AI | 72.7 | 2026-11-09 |
| 24 | GPT-6.1 Terra (unknown thinking) | OpenAI | 72.2 | 2026-10-26 |
| 25 | GPT-5.6 Sol (pro, xhigh) | OpenAI | 71.7 | 2026-07-09 |

Notes: Scores published on simple-bench.com, as mirrored by Epoch AI (simplebench_external.csv), converted to % (1 decimal). '(unknown)' means the reasoning-effort setting was not stated. Human baseline figure is from simple-bench.com.
