Press Space to continue
DeepSeek V4 evaluation reveals capability gaps vs US frontier
Press Space to continue
ARC-AGI-2 used by @NIST CAISI as flagship benchmark for DeepSeek V4 evaluation On DeepSeek’s self-reported evaluations, V4 appeared roughly competitive with frontier US models On CAISI’s broader testing suite - including ARC-AGI-2, the capability gap became much more visible
open on x · 2 viewsViews and clicks count once per visitor per day. Clicks are opens of this post on X.