Anthropic launches Claude Fable 5.2, higher costs than GPT-6 Astra

An author tested Anthropic's Claude Fable 5.2 against OpenAI's GPT-6 Astra on three developer tasks to compare costs and performance. While Fable 5.2 scored slightly higher on code review, it consumed significantly more tokens and reasoning overhead, resulting in higher overall API costs and longer processing times compared to Astra. Open article →