In short: Anthropic has informed Claude Pro and Max subscribers that access to third-party AI frameworks, starting with OpenClaw, is now restricted. Effective April 4, 2026, users will incur additional costs through a new pay-as-you-go billing tier, shifting the financial burden of running autonomous agents onto the users. The creator of OpenClaw, who joined OpenAI recently, expressed disappointment, calling this move a betrayal of open-source developers. Many users now face potential cost hikes of up to 50 times their previous monthly fees.

Anthropic has terminated a previously unannounced subsidy that positioned its Claude models as a leading choice for the open-source AI agent community. Beginning April 4, 2026, those using Claude’s Pro and Max subscription tiers are barred from utilizing their plans' usage limits with third-party frameworks like OpenClaw. To continue using these tools with Claude, users will have to pay separately under a new 'extra usage' billing system. Anthropic plans to extend this restriction to all third-party platforms shortly.

This announcement has startled thousands of developers who built their AI setups on the premise that a flat monthly subscription would suffice. For many, that is no longer the case.

The Economics Behind the Change

The rationale for the change is clear, though the timing has raised eyebrows. Claude’s subscription plans were originally designed for conversational use—where a user types a query and receives a response. However, agentic frameworks function differently. A single OpenClaw instance running autonomously throughout the day can lead to API costs ranging from $1,000 to $5,000, depending on the tasks involved. With a $200 monthly Max subscription, this places an unsustainable financial burden on Anthropic, which had been covering these costs.

“Anthropic’s subscriptions weren’t built for the usage patterns of these third-party tools,” stated Boris Cherny, Head of Claude Code at Anthropic. “Capacity is a resource we manage thoughtfully, prioritizing our customers who are using our products and API.”

The issue is substantial, with over 135,000 OpenClaw instances estimated to have been active at the time of the announcement. Analysts noted a price disparity of more than five times between what heavy agentic users were paying under flat subscriptions and what similar usage would cost at API rates. Essentially, Anthropic’s subscription model had been quietly subsidizing a usage class it had not adequately priced.

Understanding OpenClaw

OpenClaw, an open-source AI agent framework, was created by Austrian developer Peter Steinberger. Initially launched in November 2025 under the name Clawdbot, it aimed to explore the implications of equipping a large language model with persistent memory and tool access, enabling communication via messaging platforms like WhatsApp and Telegram. The response was overwhelmingly positive, indicating significant demand.

The project underwent two renamings in late January 2026, first to Moltbot, following trademark concerns from Anthropic, and then to OpenClaw shortly after. By March 2, 2026, it had garnered 247,000 stars on GitHub and 47,700 forks, with many calling it the fastest-growing GitHub project ever, achieving 100,000 stars within 48 hours at its peak. OpenClaw supports over 50 integrations and works across platforms such as Claude, GPT-4o, Gemini, and DeepSeek, with Tencent developing an enterprise solution based on it, showcasing its extensive impact beyond individual users.

Timing Concerns

The timing of this restriction is particularly notable in light of events from February. On February 14, 2026, Steinberger announced he was leaving OpenClaw to join OpenAI, where he would focus on developing the next generation of personal agents. OpenAI has committed to supporting OpenClaw as part of its open-source foundation. Steinberger indicated that partnering with OpenAI was the quickest way to expand accessibility.

Anthropic’s restrictions were implemented shortly after this transition, a timeline that has not gone unnoticed. Steinberger and investor Dave Morin sought to negotiate a more favorable outcome with Anthropic, but they only managed to postpone the enforcement by a week.

“First they copy some popular features into their closed harness, then they lock out open source,” Steinberger remarked regarding the ban.

Whether this timing is a strategic decision or mere coincidence, the result remains the same: the leading open-source agent framework now affiliated with OpenAI faces financial barriers to operating under Claude’s subscription model.

Impact on Users

The new billing structure poses a substantial challenge for developers who had relied on unlimited agentic runs under a fixed plan. The pay-as-you-go extra usage system means costs per interaction are projected to range between $0.50 and $2.00, making intensive agentic use prohibitively expensive compared to the previous flat rates. Users have reported experiencing cost increases of 10 to 50 times their former expenses, with hobbyist developers and solo practitioners particularly vulnerable.

To ease the transition, Anthropic is offering two concessions: subscribers will receive a one-time credit equivalent to their monthly plan cost, valid until April 17, and discounts of up to 30% for users who pre-purchase extra usage bundles.

Those wishing to continue using OpenClaw with Claude can do so by acquiring extra usage bundles or by obtaining a separate Claude API key, which bypasses subscription limits but charges at full API rates: $3 per million input tokens and $15 per million output tokens for Claude Sonnet 4.6 and $15 and $75 respectively for Claude Opus 4.6.

Anthropic's Ecosystem Strategy

This decision aligns with a broader trend at Anthropic. In March 2026, the company committed $100 million to its Claude Partner Network, establishing a formal network of enterprise consulting and integration partnerships centered on its products. Additionally, Anthropic has launched a marketplace for Claude-powered applications, allowing enterprise customers to purchase third-party software through channels controlled by Anthropic without taking a commission.

This approach reflects Anthropic’s intent to maintain control over customer relationships, revenue, and data. Their own developer environment, Claude Code, remains available within Pro and Max subscription plans and is not subject to the new restrictions. The underlying message for developers is clear: engage with Anthropic’s ecosystem or incur API costs for external development.

Following a $3 billion funding round in early 2026, which emphasized aspirations of developing “artificial super-intelligence for science” and expanding research capabilities, it has become evident that Anthropic is responding to the commercial pressures of managing one of the most computationally demanding products globally. As the AI industry transitions from rapid user acquisition in 2025 to understanding the actual costs of maintaining those users in 2026, this shift in strategy has significant implications for all stakeholders involved.

Source: TNW | Anthropic News