Press Space to continue
GPT 5.4 Pro Scores 38% on FrontierMath Tier 4
Press Space to continue
GPT 5.4 Pro just crushed the hardest math benchmark, FrontierMath Tier 4, with 38%. These 50 research level math problems can take mathematicians weeks to solve. A year ago, the best was 2% (o3). The best open source model is 4.2% (Kimi K2.5) Insanely impressive.
open on x · 2 viewsViews and clicks count once per visitor per day. Clicks are opens of this post on X.