Are you a subscriber of Claude Pro ($20 monthly) or Max ($100-$200 monthly) plans from Anthropic and use its Claude AI models and products to power third-party AI agents such as open claw? If so, you will be in for an unpleasant surprise.
anthropic announced a few hours ago that beginning tomorrow, Saturday, April 4, 2026, at 12 pm PT/3 pm ET, it will no longer be possible for Claude subscribers to use their subscriptions to connect Anthropic’s Claude models to third-party agent tools, citing the strain that such use was placing on Anthropic’s computing and engineering resources, and the desire to serve a large number of users reliably.
"We’ve been working hard to meet the increased demand for Claude and our subscriptions were not built for the usage patterns of these third-party tools." wrote Boris Cherny, director of Claude Code at Anthropic, in a post on X. "Capacity is a resource that we carefully manage and prioritize for our customers who use our products and APIs."
The company also supposedly sent an email in this sense to some subscribers. However, it is not certain whether Claude Team and Enterprise subscribers will be similarly affected. We’ve reached out to Anthropic for further clarification and will update you when we hear back.
To be clear, it will still be possible to use Claude models like Opus, Sonnet and Haiku to power OpenClaw and similar external agents, but users will now have to opt for a pay-as-you-go basis. "extra use" billing system or use Anthropic’s application programming interface (API), which charges for each usage token instead of allowing unlimited usage up to certain limits, as the Pro and Max plans have allowed until now.
The reason for the change: ‘third-party services are not optimized’
The technical reality, according to Anthropic, is that its proprietary tools like Claude Code, its AI vibe coding harness, and Claude Cowork, its enterprise application interface and control tool, are designed to maximize "fast cache hit rates"—reuse previously processed text to save it in computing.
Third-party harnesses like OpenClaw often overlook these efficiencies. “Third-party services are not optimized in this way, so it is very difficult for us to do it sustainably,” Cherny explained in more detail on X.
He even revealed his own practical attempts to close the gap: “I put in some PR to improve the cache hit rate for OpenClaw in particular, which should help people using it with Claude via API/overages.”
Before the news, Anthropic had also begun imposing stricter rules. Claude session limits every 5 hours. of usage during business hours (5am-11am PT/8am-2pm ET), meaning the number of tokens you could send during those sessions decreased.
This frustrated some power users who suddenly started reaching their limits much faster than before; a change that Anthropic said was to help. "manage the growing demand for Claude" and would only affect up to 7% of users at any given time.
Discounts and credits to soften the blow.
Anthropic won’t ban third-party tools entirely, but it will move them to a different ledger. the new "Additional use" The packages represent a middle ground between a flat-rate subscription and a full enterprise API account.
-
The credit: TO "soften the blow," Anthropic is offering existing subscribers a one-time credit equal to the price of their monthly plan, redeemable through April 17.
-
The discount: Users who pre-purchase "extra use" Packages can receive up to 30% discountan attempt to retain power users who might otherwise abandon.
-
Capacity management: Anthropic’s official statement noted that these tools put a "huge strain" in the systems, which forces us to prioritize "customers who use our core products and API."
‘The all-you-can-eat buffet just closed’
The response from the developer community has been a mix of analytical acceptance and marked frustration.
Growth Marketer Aakash Gupta spotted in X that he "open buffet that just closed," considering that a single OpenClaw agent running for a day could burn between $1,000 and $5,000 in API costs. “Anthropic was eating up that difference on every user that was routed through a third-party harness,” Gupta wrote. “That’s the pace of a company seeing its margin evaporate in real time.”
However, Peter Steinberger, the creator of OpenClaw who was recently hired by OpenAI, took a more skeptical view of "ability" argument.”It’s funny how times coincide”, Steinberger posted on X. “First they copy some popular features into their closed harness, then they block open source.”
In fact, Anthropic recently added some of the same capabilities that helped OpenClaw become trendy, such as the ability to message brokers through external services like Discord and Telegram — to Code Claude.
Steinberger claimed that he and fellow investor Dave Morin tried "speak sense" at Anthropic, but they were only able to delay enforcement by a week.
User @ashen_onefounder of Telaga Charity, expressed a concern likely shared by other small-scale builders: “If I switch both (OpenClaw instances) to one API key or the additional usage you recommend here, it will be too expensive to be worth using. I’ll probably have to switch to a different model at this point.”
“I know it sucks,” Cherny responded. “At its core, engineering is about trade-offs, 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 in the best way.
Licensing and the shadow of OpenAI
The timing of the crackdown is particularly notable given the migration of talent. When Steinberger joined OpenAI in February 2026, he brought the "open claw" spirit with him.
OpenAI seems to be positioning itself as another solution "harness friendly" alternative, potentially using this moment as a customer acquisition channel for disgruntled Claude power users.
Restricting subscription limits to your own "closed harness," Anthropic is asserting control over the UI/UX layer. This allows them to collect telemetry and manage rate limits more granularly, but risks alienating the power user community that built the system. "agent" ecosystem first.
The conclusion
Anthropic’s decision is a cold calculation of margins versus growth. As Cherny noted, "Capacity is a resource that we manage carefully."
In the AI landscape of 2026, the era of unlimited, subsidized computing for third-party automation is over.
For the average Claude.ai user, the experience remains unchanged; For power users running standalone offices, the bell has rung.
