28
143 forecasters
What will be the highest scores in these ΑΙ benchmarks before September 2026?
FrontierMath (Tier 4)
AGI-ARC-2
OSWorld Verified
Vals Index
GPT-5.4 claims boost in FrontierMath score
More problems for FrontierMath (Tier 4)
DeepMind's advances boost FrontierMath scores
More problems for FrontierMath (Tier 4)
Claude Opus 4.7 improves Vals index score
More % for Vals Index
GPT-5.4 Launch: OpenAI's Most Powerful Professional Model with Native Computer Use - API易文档中心docs.apiyi.com•Mar 14, 2026
Increases