Folks who rely on Claude through third party tools are about to see a change. According to a post on X by Anthropic employee Boris Cherny, Claude subscriptions will no longer cover usage inside third party applications such as OpenClaw starting Today, April 4, at 3pm ET. That means users who previously logged into tools like OpenClaw with their Claude account will now need to pay separately for the usage they generate.
In practical terms, the familiar “unlimited style” Claude subscription will only apply to the official Claude interface and services offered directly by Anthropic. Third party clients will require either usage bundles or a standard Claude API key, both of which are billed based on consumption. Anthropic says the bundles will be offered at a discounted rate, but the bottom line remains the same: heavy users who relied on automation tools could see costs climb.
The reasoning appears to come down to demand. In follow up posts, Cherny explained that Claude subscriptions were not designed for the usage patterns created by third party tools. Those apps often automate prompts, chain tasks together, or run repeated queries in the background. That kind of activity can multiply the number of model calls dramatically compared to a human chatting through the official Claude interface.
In other words, the system was built for people typing questions into a chat window. It was not built for bots, scripts, and autonomous AI workflows hammering the model around the clock.
That distinction matters because third party tools have become extremely popular with developers and AI experimenters. Projects like OpenClaw give Claude users a desktop style interface, local integrations, and automation features that the official interface does not always provide. Some people also use these tools to build AI agents that can write code, summarize documents, or run tasks continuously.
When those workflows are powered by a flat rate subscription, the economics can get messy for the company running the model. If a single user runs automated jobs that generate thousands of prompts per day, the compute cost for the provider can quickly exceed the price of the subscription.
Anthropic’s change is essentially a way to separate casual chat usage from automated workloads. If you just want to talk to Claude, your subscription still works the same way. But if you are using Claude as a backend engine inside scripts or external software, the company now wants that usage routed through the API pricing model instead.
To soften the transition, Anthropic says subscribers will receive a one time credit equal to their monthly plan cost. Users who do not like the change can also request a refund through a link that will be sent via email.
Still, the move could frustrate developers who built workflows around the assumption that their Claude subscription covered these third party integrations. Many of those tools emerged precisely because power users wanted more flexibility than the official interface offered.
It also raises a broader question about the future of “unlimited” AI subscriptions. As AI models become more powerful and more people start building automated systems around them, the gap between casual usage and programmatic usage grows wider. Companies may increasingly push heavy workloads toward API billing instead of flat rate plans.
For now, Anthropic appears to be drawing a line between chatting with Claude and running Claude as infrastructure.
Developers using third party clients will soon have to decide which side of that line they want to be on.
