Gemini 4 Carbon (high) leads Artificial Analysis’s current intelligence index, ahead of Claude Opus 5.5 (max with fallback).

In the default all-weights leaderboard observed November 4, 2026, the two variants score 60 and 59 respectively on Intelligence Index v5.0.1. Gemini 4 Carbon (medium) scores 58; Claude Opus 5.5 (xhigh with fallback) and GPT-6 Astra (max) each score 57. These are distinct reasoning-effort variants, not separate model families.

Artificial Analysis estimates weighted benchmark costs of $7.76 per task for the leading Gemini variant and $10.04 for the leading Claude variant. Gemini’s medium variant costs $6.59 per task at a displayed score of 58. These figures reflect benchmark token use and provider pricing, not a universal cost for every task.

The index combines nine evaluations across knowledge work, coding, science and general knowledge. It is primarily English-language and text-based; results do not establish superiority on every application. This is an initial snapshot, not evidence of a change from an earlier methodology.

Sources: leaderboard and intelligence methodology.