Claude Fable 5.2 costs more than Opus 5.5

An author tested Anthropic's Claude Fable 5.2 against Claude Opus 5.5 across three coding and agent tasks to determine if the cheaper model is sufficient. The tests showed that both models achieved identical scores on every version, but Fable cost nearly four times as much and took more than twice as long. Consequently, the author recommends starting with Opus 5.5 and upgrading to Fable only when necessary. Open article →