Anthropic’s new flagship model allows users to choose the level of reasoning effort devoted to each query, thereby striking a balance between cost and capability.
Anthropic made Claude Opus 5 available on July 24, 2026. The most visible new feature is not a benchmark score but a setting: the user can now choose the level of reasoning effort — low, medium, high — that the model devotes to a given task, with turn-by-turn control. In concrete terms, you can make your own choice between speed, cost and depth of analysis rather than having to make a compromise set upstream.
The model has a context window of one million tokens and a pricing of $5 per million tokens in input and $25 in output, the same price as the Opus 4.8 generation that preceded it. It is positioned as a much cheaper alternative to the company's top-of-the-range model, for a similar level of capacity.
Claude Opus 5 becomes the default model of the Claude Max offer and the most powerful option available to Pro subscribers. It is also distributed via the API and partner cloud platforms.
Read the full article on the original source
Go to the source →