Opus 5.5 edges GPT-6 Astra, 58 to 57, on Artificial Analysis

The new v5 of the Artificial Analysis Intelligence Index ranks Claude Opus 5.5 (max) first at 58, and Anthropic and OpenAI hold the top five. Read the lead as a narrow margin measured at each model’s max setting, on an index Artificial Analysis says is mostly built on private datasets.

GPT-6 Astra (max) follows at 57 and GPT-6.1 Sol (max) at 55, while Gemini 4 Argon, scored at high rather than max, lands at 50. Arena’s overall ranking puts those two OpenAI models the other way round, with GPT 6.1 Sol second and GPT 6 Astra behind it, per an earlier Threads post. Version 5 adds two new benchmarks, FrontierCode 2.0 from Cognition and Terminal-Bench-Science.