AI Benchmarks News

Latest Articles

• Frontier AI Reasoning Reaches New Heights as Grok and GPT Tie for Intelligence Crown

  Published: Apr 24, 2026

  Section: Technology

  The latest TrackingAI benchmark for April 2026 reveals a significant leap in artificial intelligence reasoning, with Grok-4.20 Expert Mode and OpenAI GPT 5.4 Pro (Vision) sharing t...

• Humanity’s Last Exam Reveals Massive Knowledge Gap Between Top AI Models and Specialized Human Expertise

  Published: Mar 13, 2026

  Section: Research

  A global coalition of 1,000 experts has launched "Humanity’s Last Exam" (HLE), a 2,500-question benchmark designed to be unsolvable by current artificial intelligence. Early result...