Artificial Analysis Intelligence Index weights shift sharply across versions, chart reveals
HamelHusain · x · 2026-11-11
Hamel Husain shows that the Artificial Analysis Intelligence Index, made up of 9 benchmarks with arbitrary weights, has shifted considerably over versions v4 through v5: τ³ Banking dropped from 14% to 5%, GDPval-AA swung between 16.7% and 20% before settling at 10%, several benchmarks were retired, and new ones like FrontierCode 2.0 and Terminal-Bench-Science entered at 10-15%. Model rankings can thus move significantly based on weight choices alone.
More from Research
- OXMAN hiring Systems Architect for EDEN, a generative ecosystem digital twin — tobias_rees · 2026-11-11
- EgoForce dataset: 2,000 hours of egocentric video paired with calibrated tactile forces — yining_hong · 2026-11-11
- JetBrains Survey of 18K Go Devs: 86% Use CLI Coding Agents Weekly — rseroter · 2026-11-11
- SNARKs for LLM inference 'finally starting to make sense,' says ZK researcher — DimitrisPapail · 2026-11-11
- Cognitive scientists find AI generates ideas faster, but human teams stay more creative — CurieuxExplorer · 2026-11-11
- AI agents ran a biology lab end-to-end: AutoDNA completes DNA read-write cycle in 200 hours — CurieuxExplorer · 2026-11-11