Anthropic has reversed course on a billing change for its Claude Agent SDK just before the change was set to take effect, according to The Decoder.
The company had planned to charge for use of the Agent SDK and third-party apps through a separate pool of credits, rather than counting that usage against a customer's existing subscription limits. The Decoder reports the plan proved unpopular and was pulled back before launch.
Under the reversal, the Agent SDK and third-party applications will continue to draw from regular subscription limits, as they did before. In practical terms, that means developers and subscribers who build on or use these tools won't have to buy and manage a second, separate set of credits to keep working.
The Decoder frames the retreat against the backdrop of an intensifying price war with OpenAI. With both companies competing aggressively on cost, a move that effectively raised the price or complexity of using Anthropic's developer tools risked pushing customers toward rivals.
The Claude Agent SDK is the toolkit developers use to build applications powered by Anthropic's Claude models, so changes to how it is billed ripple outward to the products and services built on top of it.
Why it matters: The reversal is a small but telling sign that competition between AI providers is now shaping not just headline model prices but the fine print of how developers get charged — and that customer pushback can move a major lab to change its plans before they even ship.