Key Points
- Anthropic paused its plan to move Claude Agent SDK usage to a separate credit pool.
- The change was set to take effect June 15 but will not proceed for now.
- Agent SDK, claude -p, and third-party app usage stay on subscription limits.
- Anthropic says it will give advance notice before any future change.
Anthropic has hit pause. In an email to subscribers on June 15 — the very day the change was set to take effect — the company said it is "not making this change today" and is "working to update the plan to better support how users build with Claude subscriptions." For now, nothing changes: Agent SDK and claude -p usage keeps drawing on existing subscription limits, with no separate credit to claim.
What was supposed to happen
The plan, announced in May, would have pulled the Claude Agent SDK, the headless claude -p command, Claude Code GitHub Actions, and third-party apps built on the Agent SDK out of subscription rate limits and onto a dedicated monthly credit — $20 for Pro, $100 for Max 5x, $200 for Max 20x — metered at full API rates with no rollover. Interactive Claude Code and chat would have stayed inside the subscription.
Why developers pushed back
For heavy users, that was a steep increase. Agent builders had been pulling hundreds of dollars of API-equivalent compute out of a $20-to-$200 subscription, and the new credit would have metered every call at list price. Developers who had built tools and workflows on that math objected loudly, and Anthropic blinked.
The pattern
This is the second time in six months Anthropic has floated a crackdown on programmatic usage and retreated. In January, it blocked subscription OAuth tokens from working with third-party tools, then reversed course within days after a similar backlash. The throughline is the same tension pulling at every lab monetizing the economics of frontier models: an interactive user sends dozens of prompts a day, while an autonomous agent can fire thousands and call models recursively.
The read
Anthropic keeps colliding with the same wall. Autonomous agents make flat-rate subscriptions hard to sustain, but the developer ecosystem has been built on exactly that subsidy, so every attempt to claw it back triggers a revolt. The pause is a concession, not a reversal of the underlying problem — and the closing line of the email, a promise to "share it with advance notice before it takes effect," is the tell. Anthropic isn't abandoning the change. It's buying time to make it land softer.
Source: Anthropic (email to subscribers), The New Stack, TechTimes
Comments