Anthropic's latest AI release has outperformed a competing model in an aggregated test of intelligence. Claude Opus 5.5 claimed first place on the Epoch Capabilities Index (ECI) with a score of 167, narrowly beating the GPT-6 Astra model from OpenAI.

GPT-6 Astra retains a lead in mathematics with 170 points compared to 168 for Opus, but Opus leads in software engineering at 167 to 165. Anthropic's mid-size model, Claude Sonnet 5.5, roughly matched the score of 165 seen in the larger Claude Fable 5.1. Index data show Opus and Sonnet 5.5 lose <1 ECI point when moving from general tasks to software engineering, while Astra's score drops by about 2 points.