Claude Fable 5.1 tops AI index but costs 20% more per task (artificialanalysis.ai)

Claude Fable 5.1 has officially topped the Artificial Analysis Intelligence Index, achieving an impressive score of 66, surpassing its predecessors and competing models. Despite a significant 75% reduction in cache read prices, it comes with a 20% increase in cost per task compared to Fable 5, resulting in a charge of $3.76 for each task on the Index due to an increase in output token usage. This enhanced version not only excels across various benchmarks—including scoring 91.4% on Terminal-Bench v2.1—but also exhibits remarkable performance in agentic tasks, where it leads on GDPval-AA v2 with an Elo score of 1,853, although it's closely matched with Claude Opus 5 in several areas. The significance of this update lies in its potential to redefine the cost-benefit analysis for deploying AI models, especially in high-stakes environments. The notable improvements in caching efficiency could alleviate some operational costs, despite the increased base pricing. With a context window of 1 million tokens and strong performance metrics across various task scenarios, Fable 5.1 positions itself as a frontrunner in the rapidly evolving AI landscape, setting a new standard for performance in relation to cost-effectiveness.