Anthropic has closed the door on its subscriptions being used for tools like OpenClaw.
Starting Sunday, April 5 at 12pm PT, Claude’s Pro and Max plans will no longer cover usage on third-party agent harnesses. Users who want to keep running OpenClaw or similar tools through their Claude login will need to purchase separate “extra usage” bundles — now available at a discount — or switch to a Claude API key with pay-as-you-go billing.
The announcement came via Boris Cherny, Head of Claude Code at Anthropic, in a post on X on Friday night. “We’ve been working hard to meet the increase in demand for Claude, and our subscriptions weren’t built for the usage patterns of these third-party tools,” Cherny wrote. “Capacity is a resource we manage thoughtfully and we are prioritizing our customers using our products and API.”
The timing is notable. Claude had a breakout few weeks — briefly topping the US Apple App Store in March — and Anthropic had already begun tightening session limits during peak hours, citing surging demand. The move against third-party harnesses is the most aggressive step yet.
The Engineering Argument
Anthropic’s case is rooted in how its systems are built. Its own tools — Claude Code and Claude Cowork — are optimized for high prompt cache hit rates, meaning previously processed text gets reused to reduce compute load. Third-party harnesses like OpenClaw largely bypass these efficiencies, consuming significantly more resources per session than first-party usage.
In follow-up comments on X, Cherny was candid about the tradeoffs: “Fundamentally engineering is about tradeoffs, and one of the things we do to serve a lot of customers is optimize the way subscriptions work to serve as many people as possible with the best model. Third party services are not optimized in this way, so it’s really hard for us to do sustainably.”
In an effort to soften the blow, Cherny said he personally submitted pull requests to improve prompt cache efficiency in OpenClaw — meaning API users and those on the new extra usage billing will at least see some cost benefits from his work.
Affected subscribers get a one-time credit equal to their monthly plan cost. Those who want a full refund can request one via a link being sent by email.
OpenClaw in the Crossfire
OpenClaw started as a weekend project by Austrian developer Peter Steinberger in late 2025. Users could text it on Telegram or WhatsApp and have it manage their calendar, triage email, run scripts, and browse the web. It gained 25,000 GitHub stars in a single day and quickly became one of the most talked-about AI tools around. It was subsequently acquired by OpenAI.
Steinberger was blunt on X: “Both me and @davemorin tried to talk sense into Anthropic, best we managed was delaying this for a week. Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source.”
The observation isn’t without merit. Anthropic recently launched Dispatch, a feature that lets users control their computer via Claude from their phone — functionality that closely mirrors what made OpenClaw popular in the first place. Anthropic’s move to cut OpenClaw’s subscription access while rolling out competing features will strike many as a deliberate competitive play, even if the company frames it as an engineering necessity.
Steinberger did give Cherny partial credit for submitting PRs to improve prompt cache efficiency in OpenClaw. “While I think what Anthropic does is sad for the ecosystem, I wanna give Boris credit for doing what he can to soften the fallout.”
Not a One-Off
This is not the first time Anthropic has drawn a hard line around third-party access to its models. In June 2025, Anthropic cut off Windsurf’s access to Claude — the coding IDE that was being acquired by OpenAI — with less than a week’s notice. And Anthropic is not alone in this pattern: Google’s Antigravity IDE similarly restricted OpenClaw access earlier this year, citing malicious misuse degrading service quality for legitimate users.
What’s emerging is a clear industry posture: AI labs built their ecosystems on openness when it served growth, and are now walling off third-party access as competitive stakes rise. Anthropic’s subscription terms already prohibited using Claude Pro or Max credentials in third-party tools — today’s enforcement simply makes that prohibition operational.
For developers who built workflows around a flat-rate Claude subscription powering OpenClaw, the options are now materially more expensive. Pay-as-you-go API billing charges per token, and even discounted extra usage bundles represent a structural cost increase over the all-you-can-use access many had come to rely on. Some users have already said publicly they will move to cheaper open-source alternatives rather than absorb the new costs.
Anthropic’s closing line from Cherny strikes a measured tone: “We want to be intentional in managing our growth to continue to serve our customers sustainably long-term. This change is a step toward that.”
Whether the developer community reads it that way is another matter.