Anthropic pauses token billing change for Claude Agent SDK

Anthropic paused a planned change that would have moved Claude Agent SDK usage onto API-style token billing, saying "for now, nothing has changed" and allowing Agent SDK activity to continue under existing Claude subscription limits. The move reverses a May 13 plan that would have billed outside SDK usage at API rates starting June 15.

What did Anthropic announce and what was paused?

Anthropic announced on May 13 that it would treat Claude Agent SDK usage separately from standard Claude chat usage, and that, as of June 15, outside SDK usage would be billed at Anthropic's prevailing API rates with subscribers receiving a monthly usage credit equal to their subscription price. On the day the change was set to take effect, Anthropic updated its billing support page to say it was "pausing the changes to Claude Agent SDK usage described below," adding that "for now, nothing has changed" while it works to update the plan to better support how users build with Claude subscriptions.

How would the billing change have affected Claude subscribers?

The May 13 plan would have capped how much heavy users could rely on subscription tiers by shifting Agent SDK calls to API pricing, replacing the current setup where Agent SDK use is limited only by the standard weekly caps of a user's Claude subscription tier. Under the existing arrangement, users can extract much more value from subscriptions: one analysis cited in the announcement suggests Claude Opus users start saving after just two to three messages per day, and that an Opus subscription could be worth many multiples of its monthly cost in API usage. Developer Matthew Diakonov wrote that "if you are a developer using Claude as your primary coding assistant with Opus, you will blow past breakeven in the first week."

What context led Anthropic to reconsider?

Anthropic's brief pullback follows other recent shifts in token-based billing across the industry, including GitHub Copilot's token-based changes that surprised many users. The company is also preparing for a possible public offering, having filed confidential papers with the Securities and Exchange Commission. Internally, Anthropic staff framed the issue as one of capacity and unplanned usage patterns: in April, Anthropic Head of Claude Code Boris Cherny said, "our subscriptions weren’t built for the usage patterns of these third-party tools," and said capacity is a resource the company manages thoughtfully as it prioritizes customers using its products and API.

Why it matters

The pause preserves a tacit subsidy for heavy Agent SDK users who had been relying on subscription caps rather than per-token API charges. For developers and third-party apps that route automation and agent traffic through subscriptions, the May-to-June change threatened to convert low incremental cost into significant monthly API bills. Anthropic's decision to step back means those applications can keep current usage patterns for the moment, but the company explicitly signaled it intends to redesign billing so subscriptions and third-party agent usage align with its capacity priorities.

What to watch

Look for a revised billing plan from Anthropic that replaces the paused May 13 proposal; the company said it is "working to update the plan to better support how users build with Claude subscriptions." The concrete signals to track are any new effective dates, the exact structure of usage credits tied to subscription price, and whether Anthropic defines different caps for third-party agent traffic or publishes new API-equivalent rate tables for Agent SDK calls.

Anthropic's short-term pause buys developers time, but not permanence. Expect commercial pressure from heavy-agent customers and operational pressure from Anthropic's capacity management to shape whatever billing rules arrive next.