VISTA Gives Multimodal Models Visual Memory
Kaiming He’s team introduced VISTA, a visual-native harness that gives existing multimodal models direct observation, lossless frame storage, and on-demand inspection without retraining the base model. The paper reports that Claude Opus 5 scored 100 and GPT-5.6 Sol scored 99 on ARC-AGI-3’s 25 public games, with improvements also reported on GameWorld, AI GameStore, and BabyVision. These are author-reported results on public tasks and do not establish generalization to private or unseen tests.
VISTA shows that substantial interactive-reasoning gains can come from changing how a model observes and retrieves visual information rather than changing its weights.
编者按:本文来自微信公众号 “新智元”(ID:AI_era),作者:ASI启示录,创业邦经授权转载。
何恺明团队的新论文,在X上炸了!
十一期间刚挂上arXiv的VISTA论文显示,过去半年,全世界最顶尖的大模型,全是被「蒙着眼睛」赶上AGI考场的。
同一个Claude Opus 5,在官方标准测试里只拿了可怜的40分。
何恺明团队只不过帮它摘掉了眼罩,让它亲眼看游戏画面,顺手递给它一本能随时往回翻的「相册」。
结果,Claude直接把ARC-AGI-3的25个公开游戏全部打穿,拿下100分满分!
#VISTA#MIT#Claude Opus 5#GPT-5.6 Sol#ARC-AGI-3#GameWorld#AI GameStore#BabyVision