Elorian 发布视觉推理模型 Glimpse 与 DecoyCount 基准
原标题：Can Frontier Models See Past a Decoy? Elorian Tests Visual Reasoning
AI 摘要
Elorian 发布首个视觉推理模型 Glimpse，并联合 Mercor 推出 DecoyCount 基准，测试模型是真正计数物体还是依赖视觉捷径。论文显示前沿模型在该基准上准确率均低于 50%，最高为 GPT-6 Astra 的 43.17%，而 Glimpse 内部测试得分为 50.48%。Elorian 由前 Google DeepMind 研究员 Andrew Dai 联合创立，主张让图像生成与编辑成为推理链的一部分，区别于 World Labs Atlas 等世界模型。
正文节选
This month Elorian launched its first visual reasoning model, Glimpse. It’s being promoted as “a new category of visual thinking model: a system that reasons about the physical world from what it sees.” So how does Glimpse compare to both frontier models and world models in vision tests? A new benchmark called DecoyCount, co-released by Elorian, provides some answers. We quizzed CEO Andrew Dai on the benchmark, and also talked about Elorian’s positioning in a crowded models landscape. DecoyCount