Claude Opus 5 Launches With Frontier-Level Performance

Anthropic has released Claude Opus 5, a high-performance model offering near-frontier intelligence at half the price of Claude Fable 5. It sets new benchmarks on coding and knowledge work evaluations including Frontier-Bench and GDPval-AA, and becomes the default model on Claude Max and the strongest on Claude Pro. Opus 5 doubles its predecessor's performance at lower cost, excels at software engineering, scientific research, and computer use tasks, and outperforms competitors on problem-solving benchmarks like ARC-AGI 3 and Zapier AutomationBench. Real-world testing shows the model autonomously solving complex engineering challenges that rival models could not complete.