Anthropic releases lower-cost Claude Opus 5 AI model

Anthropic released Claude Opus 5, a new artificial intelligence model that the company said comes close to the performance of its strongest publicly available system at half the price.

The San Francisco-based company said the model leads industry benchmarks for coding and knowledge work while using fewer computing resources than earlier versions. The company linked the launch to concerns from business customers about the cost of running advanced AI systems.

Claude competes with OpenAI's ChatGPT and Google's Gemini. Customers pay according to the amount of work performed, measured in tokens. Anthropic said Opus 5 costs $5 per million input tokens and $25 per million output tokens. It said that is half the price of Fable 5, its most expensive publicly available model, while Opus 5 performs close to Fable 5 across many tasks.

Cost has become a major issue after Anthropic and OpenAI filed for stock market listings in June. The source said Wall Street is examining their business performance more closely before their market debuts.

On cybersecurity, Anthropic said it deliberately avoided training Opus 5 on cyber tasks. It said Opus 5 remains behind Mythos 5, its most capable and most tightly restricted system, on both offensive cyber work and biological research capabilities that could be misused. Mythos 5 is available only to selected partners.

Anthropic said Opus 5 finds software flaws about as well as Mythos 5 but is far less capable of turning those flaws into attacks. The company said that, because of the reduced risk, its automated safety filters will intervene about 85 percent less often on Opus 5 than on Fable 5, a Mythos-class model released last month.

Fable 5 and Mythos 5 were briefly removed from public access in June after the US Commerce Department imposed export controls. The source said those controls were lifted at the end of that month.