benchgap
Calibration

BabyVision w/ Python → PerceptionBench

PerceptionBench is estimated from BabyVision w/ Python with a inverse Michaelis–Menten curve fitted on 5 models measured on both: y = 0.74680·(x − 0.0000) / (0.0000 + 2.0000 − x), R² = 0.84, cross-validated error 2.5 pp. It is used for 4 estimates.

Estimated modelBabyVision w/ PythonPerceptionBenchSource
Glimpse95.4%68.1%estimated ± 2.5 pp, medium confidence
DeepSeek V4.1 Flash 101792.5%64.3%estimated ± 2.5 pp, medium confidence
MiniMax M3.1 Flash91.0%62.3%estimated ± 2.5 pp, medium confidence
Qwen3.8-27B85.6%55.9%estimated ± 2.5 pp, low confidence

Fitted on

ModelBabyVision w/ PythonPerceptionBench
Inkling 2 Mini88.8%57.5%
Inkling 2 Small93.7%64.7%
Kimi K3.1 Pro95.4%68.3%
Kimi K385.7%58.5%
Qwen3.8 Max91.3%63.5%