Anthropic launches Claude Opus 5.5, says operating cost is lower

Anthropic launched Claude Opus 5.5 on September 22. Reuters reported that the company says Opus 5.5 delivers performance comparable to its top-tier Fable 5.1 while costing about 40 percent less to run than its predecessor. Anthropic also said Opus 5.5 outperformed OpenAI GPT-5.6 Sol on a software-development benchmark. The comparison comes from a company-reported benchmark. The release arrives amid continuing debate over the pace, cost, capability and safety of advanced artificial-intelligence systems. Lower inference costs can matter to businesses and developers because advanced models may be used repeatedly across coding, research and automated workflows. Benchmark results can vary by task and testing method. The launch adds another model to a fast-moving market in which developers are competing on capability, price, reliability and safety.