Tag
Anthropic released Claude Opus 4.8, claiming it is the only model to complete every case on the Super-Agent benchmark and that it outperforms GPT-5.5 on browser/computer use tasks with better tool efficiency and fewer uncorrected code flaws.