Benchmarks · reasoning
BabyVision
UniPat AI's visual-reasoning test of basic perception a young child handles (tracking, pattern completion, visual puzzles): 388 items, 22 task types. Accuracy without tools; humans score 94.1%.
11 models · 11 results · data as of 2026-10-29 · unipat.ai ↗
Leaderboard
Qwen 3.8 Max max
All results
11 rows| # | Model | Effort | BabyVision |
|---|---|---|---|
| 1 | Qwen 3.8 Max | max | 82.0% |
| 2 | Gemini 3.1 Pro | high | 51.6% |
| 3 | Gemini 3.0 Pro | 49.7% | |
| 4 | GPT-5.4 | xhigh | 49.7% |
| 5 | Qwen 3.5 397B-A17B | max | 43.3% |
| 6 | Kimi K2.5 | max | 36.5% |
| 7 | GPT-5.2 | xhigh | 34.4% |
| 8 | Qwen 3 | high | 19.2% |
| 9 | GLM 4.6 V | low | 17.6% |
| 10 | Grok 4 | high | 16.2% |
| 11 | Opus 4.6 | max | 14.8% |