Claude Fable 5.2 tops Artificial Analysis index at 65, far above rivals on cost curve
YashasGunderia · x · 2026-11-10
The Artificial Analysis Intelligence Index v5.0.1 shows Claude Fable 5.2 (max) leading at 65, ahead of Fable 5.2 (high) variants at 62, Google Gemini 4 Carbon at 60, and Claude Opus 5.5 at 59 — though several top entries have no publicly available API.
A companion cost-vs-intelligence chart plots release lines by vendor: Anthropic models occupy the highest-intelligence region (Fable 5.2 around $30-50/task), while Kimi K3.1 Pro, DeepSeek, GLM, and MiniMax models cluster in the sub-$1-per-task region.