Artificial Analysis breaks down every evaluation in Intelligence Index v5

ArtificialAnlys · x · 2026-11-18

Artificial Analysis publishes the per-evaluation breakdown of Intelligence Index v5, covering AA-Briefcase v1.1 (Elo), AA-Omniscience v1.1 (knowledge & hallucination), AutomationBench-AA, Terminal-Bench-Science 0.1, FrontierCode-AA 2.0, HLE-Diamond, GDP.pdf, Terminal-Bench 4.0 and CriPt v2. Claude Fable 5.2 (max) leads many charts while GPT-6.2 models top Terminal-Bench-Science and AA-Omniscience accuracy.

Original post →

More from Models

Models channel →