Claude Opus 5 tops agentic benchmarks and costs 26% less than rival Fable 5

Anthropic has released Claude Opus 5, which has claimed the top position for agentic knowledge work according to benchmarks published by Artificial Analysis, a firm that supported the model's pre-release evaluation. The model scored 1861 Elo on GDPval-AA v2, more than 100 points ahead of both Fable 5 and GPT-5.6 Sol, and achieved joint first place on the Artificial Analysis Coding Index. At maximum effort, Opus 5 costs $2.03 per task compared to Fable 5's $2.75, representing a 26% reduction for equivalent or superior agentic performance. The model carries a 1 million token context window, five effort settings, and is priced at $5 per million input tokens and $25 per million output tokens. However, its hallucination rate rose 14 percentage points to 50% over its predecessor, making it less suitable for tasks requiring factual precision, where Fable 5 retains an advantage.