Latent Space· Richard MacManus·· 18 d ago
Can Frontier Models See Past a Decoy? Elorian Tests Visual Reasoning
Can Frontier Models See Past a Decoy? Elorian Tests Visual Reasoning
AI summary
A new decoy counting benchmark exposes a gap between seeing and reasoning. Elorian’s Glimpse tries to bridge it by thinking with images.
Selection record
Threshold 76Media and individualsFirst 47Second 47
Not admittedSum of both 94 < twice the threshold 152
- Source tier
- Media and individuals; this tier's threshold is 76
- Pre-filter
- passed：视觉推理模型与基准测试
- Same event
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Latent Space · latent.space