Opus 5.5 still tops AA intelligence, but GPT-6.1 Sol shifts the cost frontier

rhiever · reddit · 2026-09-30

A Reddit user compared reasoning-effort settings across models using Artificial Analysis's Intelligence Index and per-task benchmark costs. Opus 5.5 still leads on intelligence, but GPT-6.1 Sol moves the cost frontier, offering far better price-performance at comparable capability levels. Costs include caching and reasoning tokens, not subscription prices. Source: artificialanalysis.ai.

Related event: GPT-6.1 Sol Redraws the Cost-Performance Frontier as Opus 5.5 Keeps the Intelligence Crown(6 posts)→

Original post →

More from Models

Models channel →