Opus 5.5 outperforms on accuracy and automation; GPT-6 Sol is up to 6.5 times cheaper

According to Artificial Analysis, Opus 5.5 scores 58 against 48 for GPT-6 Sol. At maximum effort, a task costs US$5.98 with Opus 5.5 and US$1.06 with GPT-6 Sol. In AutomationBench, Opus 5.5 achieves 40% successful tasks against 33.2%.