Anthropic Is Forcing Users to Pay Extra to Run OpenClaw With Claude

Bad news, OpenClaw fans: Anthropic wants you to pay more to use its AI models. This wasn't something Anthropic necessarily announced, either; rather, the company started sending emails to affected users, letting them know they could no longer use their Claude Code subscription limits with third-party "harnesses," including—and most notabl—OpenClaw. Anthropic confirmed users could still connect to OpenClaw with their Claude account, but they'd have to pay more money in order to do so—including a "pay-as-you-go" option tacked onto the cost of the subscription.

According to Anthropic, this policy change isn't without logic or reason. As highlighted by TechCrunch, Boris Cherny, head of Claude Code, explained on X that the company's subscriptions "weren't built for the usage patterns of these third-party tools," and that Anthropic is prioritizing customers who are using the company's first-party products and API.

OpenClaw has had quite a ride. The tool, which was previously called Moltbot, and first called Clawdbot, is designed to be an agentic AI assistant you run locally on your devices. For many, that means running OpenClaw on a Mac mini designed for this one purpose. Unlike ChatGPT or Gemini, which has their own proprietary interfaces, you communicate with OpenClaw through any chat app of your choice. You can text OpenClaw in WhatsApp, Apple Messages, Microsoft Teams, whatever you want, to organize your email inbox, write code for a project, plan out your goals for the month, whatever it is you want your agent to do.

But OpenClaw doesn't just...run. You need to power it with an AI model. In this case, users are relying on Anthropic's Claude—and, if they had a Claude Code subscription, they were simply tapping into that plan they already paid for. As you might expect, running agentic tasks through OpenClaw is extremely intensive, which pushed Anthropic to rethink how it was charging users.

Interestingly, OpenClaw's founder, Peter Steinberger, joined OpenAI back in February—one of Anthropic's chief rivals. Steinberger said on X that he and OpenClaw board member Dave Morin "tried to talk sense into Anthropic, [but] best we managed was delaying this for a week."