Artificial Analysis releases Intelligence Index v5 with new coding and science benchmarks

Artificial Analysis has released version 5 of its Intelligence Index, a composite measure of AI model capabilities. The update introduces two new benchmarks focused on coding and science tasks, plus three upgrades to existing measures. For the first time, the majority of the index is built on private datasets rather than public ones, designed to better reflect real-world use cases.

Claude Opus 5.5 ranks at the top of the updated index at a score of 58. The shift to private datasets is intended to provide a more comprehensive view of how models perform beyond standard public benchmarks, which can become saturated or gamed over time.