Benchmarks / Reasoning
Humanity's Sixth Sense
Scale AI and Elorian's intuitive visual reasoning benchmark: 522 image and silent-video tasks covering temporal, spatial, social and abstract inference. Pass@1 averaged over three attempts per task; all rubric criteria must pass. Published intervals are 95% task-cluster bootstrap intervals.
Humanity's Sixth Sense
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Gemini 4 | high | 55.4% |
| 2 | GPT-6 Astra | max | 53.6% |
| 3 | GPT-6.1 Sol | max | 46.6% |
| 4 | Opus 5.5 | xhigh | 44.6% |
| 5 | Gemini 3.8 Flash | high | 41.6% |
| 6 | Fable 5.1 | max | 40.8% |
| 7 | Gemini 3.7 Flash | high | 39.8% |
| 8 | Muse Spark 1.3 | max | 37.4% |
| 9 | Fable 5 | max | 34.5% |
| 10 | Qwen 3.8 Max | max | 33.0% |
| 11 | Gemini 3.5 Flash | high | 32.6% |
| 12 | Gemini 3.6 Flash | high | 31.9% |
| 13 | GPT-6 Sol | max | 31.2% |
| 14 | Qwen 3.8 Flash | max | 30.9% |
| 15 | Opus 5 | max | 30.5% |
| 16 | GPT-5.6 Sol | max | 30.0% |
| 17 | GLM 5.3 Flash | max | 26.8% |
| 18 | Kimi K3 | max | 25.5% |
| 19 | MiMo V2.6 Flash | high | 25.4% |
| 20 | MiMo V2.6 Pro | high | 25.2% |
| 21 | Sonnet 5 | max | 24.8% |
| 22 | Muse Spark 1.2 | xhigh | 24.5% |
| 23 | Opus 4.8 | max | 23.5% |
| 24 | GPT-5.6 Terra | max | 22.7% |
| 25 | GPT-5.6 Luna | max | 21.6% |
| 26 | GPT-6 Luna | max | 21.0% |