Artificial Analysis Releases Intelligence Index v5 With 60% Private Benchmarks; Claude Opus 5.5 Tops Rankings

Artificial Analysis has launched Intelligence Index v5, its most comprehensive benchmarking overhaul to date, addressing a critical vulnerability in AI evaluation: benchmark contamination. For the first time, 60% of the index is built on private test sets or held-out answer keys, a direct response to the problem of public benchmark data leaking into model training sets and inflating scores.