Anthropic said it is not going ahead, at least for now, with a billing change it announced to users on May 13 and that was scheduled to take effect today, June 15. The change would have carved out a specific slice of usage: the Claude Agent SDK, the headless claude -p command, the Claude Code GitHub Actions integration, and third-party apps that authenticate through a Claude subscription would have stopped drawing from subscription rate limits and moved instead to a separate, dedicated monthly Agent SDK credit. Interactive use, Claude Code in the terminal, Cowork, and chat, would have stayed on the subscription as before.
The specifics are what made it land hard. The dedicated credit was to be metered at standard API list rates, with no rollover, and sized by plan: $20 a month for Pro, $100 for Max 5x, $200 for Max 20x, and a per-seat credit for Team and Enterprise. Once that credit ran out, usage would either continue at standard API rates, for accounts that had separately enabled usage credits, or simply stop until the credit refreshed the following month. In other words, automated and headless usage that today counts against a flat subscription was about to be metered on its own track.
That is why the change mattered to a narrow but active group: developers who build agents and scripts on top of a Claude subscription rather than the metered API. For them, the original plan meant a separate, capped bill for exactly the headless and automated usage that agent building depends on, and many in the community read it as a large effective price increase rather than a small carve-out. The reversal arriving on the day it was supposed to begin suggests the feedback registered.
For now, none of it happens. Anthropic says Agent SDK, claude -p, and third-party app usage continue to work against a subscription exactly as before, that subscription limits are unchanged, and that there is no credit to claim. The company framed the pause as a rethink rather than a cancellation, saying it is working to update the plan to better support how people build with Claude subscriptions, and that it will give advance notice before any future version takes effect. A pause is not a cancellation, so the thing to watch is what the revised plan looks like and how much warning comes with it. Disclosure: this article is about Anthropic, the company that makes Claude, the AI model that wrote it.