Why did Anthropic cut Claude usage on OpenClaw?

Anthropic is changing how Claude subscriptions can be used with third-party tools like OpenClaw. Starting April 4 at 12pm PT, Claude subscriptions will no longer cover usage on OpenClaw, a move Anthropic frames as a way to better manage compute capacity. The policy effectively forces developers and teams that rely on OpenClaw to pay additional costs if they want continued access to Claude-powered agent workflows.

The change matters because it targets one of the most visible "agentic" development ecosystems in the market: OpenClaw and similar third-party agents that route tasks to LLMs through external tooling. By decoupling third-party tool usage from included subscription benefits, Anthropic is tightening the relationship between its paid offerings and where compute is consumed.

Operationally, the restriction can affect:

At least one additional related story indicates the restriction is also being implemented in a way that makes usage more expensive for subscribers rather than fully blocked outright, suggesting Anthropic may offer add-on credit or a mechanism to continue using Claude via these tools at higher marginal rates.

Overall, the policy signals that LLM providers are increasingly treating third-party agent ecosystems as a capacity-management problem -- not just an integrations problem. As more AI tools become "agent-first," subscription plans are likely to be rewritten around usage controls, capacity limits, and metering, rather than flat entitlements.