【AI前沿】Anthropic Urgently Halts Claude Agent SDK Billing Change, Heavy Users Temporarily Maintain Existing Subscription Limits

AI NEWSLatest AI NewsArticleAnthropic Urgently Halts Claude Agent SDK Billing Change, Heavy Users Temporarily Maintain Existing Subscription LimitsPublished in Latest AI NewsTime :Jun 17, 2026Read :3minuteArtificial intelligence startup Anthropic suddenly announced on Monday that it has paused its planned June 15 implementation of the Claude Agent SDK billing plan adjustment. The plan originally aimed to distinguish the use of external SDKs through third-party applications and programmatic “claude-p” commands from the standard web and official CLI chat interfaces, instead fully integrating with the current API billing rates (subscribers only receive equivalent monthly credit systems).This move means that the pricing for this intelligent agent development tool will remain unchanged temporarily, allowing heavy users and developers to continue enjoying the relatively more flexible weekly standard limits in their current Claude subscription levels.Previously, the flexible subscription mechanism led to high user engagement among advanced developers. Analysis indicated that Claude Opus users could achieve a break-even point by sending only two to three messages per day through this model. Therefore, the originally planned token-based billing change had caused strong cost concerns among developers of code editors such as Zed.Although Anthropic’s official statement said this “status quo maintenance” was to further optimize the plan to better support the development needs of subscribers, it created tension with Boris Cherny, the head of Claude Code at Anthropic, who previously stated that the subscription model was not designed for third-party automated proxy tools.Amid the recent introduction of similar token-based billing by GitHub Copilot, and reports that Anthropic is preparing to secretly file an IPO document with the U.S. Securities and Exchange Commission, this sudden halt in the billing strategy reflects the complex trade-offs that large model manufacturers face between commercialization pressure, computing capacity management, and developer ecosystem stickiness.