#Benchmarks

This topic hub collects our latest reporting and commentary on Benchmarks. Use it to follow new developments, revisit older angles, and jump between related market stories from one archive.

• AGI Hype vs. Reality: The Gap Between Claims and Capability

  19 May, 2026

  The ARC-AGI-3 benchmark reveals top AI models score under 1% on true generalization tasks, exposing the gap between AGI hype and reality.