Anthropic restricts OpenClaw from accessing Claude subscriptions as part of its cost-cutting measures | TNW

      In summary: Anthropic has restricted Claude Pro and Max subscribers from utilizing their flat-rate plans with third-party AI agent frameworks, starting with OpenClaw. This change, effective from April 4, 2026, transfers the costs of operating autonomous agents to users via a pay-as-you-go billing model. The creator of OpenClaw, who joined OpenAI in February, labeled this decision a betrayal of open-source developers. Consequently, thousands of users are now confronted with cost increases of up to 50 times their previous monthly expenses.

      Anthropic has concluded a discreet subsidy that had positioned its Claude models as the preferred choice for the open-source AI agent community. As of April 4, 2026, users subscribed to Claude’s Pro and Max tiers can no longer use their plans with third-party frameworks like OpenClaw. To continue using these tools with Claude, they must now pay separately under an "extra usage" billing system. Anthropic plans to extend this restriction to all third-party platforms shortly.

      This announcement shocked numerous developers who had based their AI setups on the idea that a flat monthly subscription was adequate. For many, this is no longer the case.

      The economic reasoning behind this shift is clear, although the timing may seem puzzling. Claude’s subscription plans were intended for conversational use: a user starts a chat, types a question, and expects a reply. However, agent frameworks function under a fundamentally different model. A single OpenClaw instance operating autonomously for a day, browsing the web, managing calendars, answering messages, and running code, can incur API costs ranging from $1,000 to $5,000, depending on the workload. Under a $200-per-month Max subscription, this creates an unsustainable shift of compute expenses from users to Anthropic.

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

      The scale of the issue was considerable. More than 135,000 OpenClaw instances were estimated to be active at the time of the announcement, and industry analysts highlighted a price gap of over five times between what heavy agent users paid under flat subscriptions and the equivalent costs at API rates. Anthropic’s subscription model had, in effect, been quietly subsidizing usage that it hadn’t accurately priced.

      What OpenClaw is and its significance

      OpenClaw is an open-source AI agent framework developed by Austrian programmer Peter Steinberger. Initially released in November 2025 under the name Clawdbot, it was initially a side project: Steinberger was curious about the results of providing a large language model with persistent memory, tool access, and the ability to communicate via messaging apps like WhatsApp and Telegram. It turned out there was significant demand for such capabilities.

      The project underwent two name changes within three days in late January 2026: first to Moltbot after Anthropic raised trademark issues regarding its similarity to “Claude,” and then to OpenClaw shortly after. By March 2, 2026, the repository had garnered 247,000 GitHub stars and 47,700 forks. It had rapidly become what many observers considered the fastest-growing GitHub project ever, reaching 100,000 stars in less than 48 hours at its peak. The framework supports over 50 integrations and is compatible with Claude, GPT-4o, Gemini, and DeepSeek. An enterprise platform built by Tencent on top of it has shown that OpenClaw’s reach extends well beyond individual developers.

      A timeliness issue

      The restriction is particularly notable given events from February. On February 14, 2026, Steinberger announced he was leaving his project to join OpenAI. Sam Altman publicly stated that Steinberger would “drive the next generation of personal agents” at the company, and that OpenClaw would transition to an open-source foundation with OpenAI’s ongoing support. In a blog post, Steinberger mentioned that “partnering with OpenAI is the most efficient way to make this available to all.”

      Anthropic implemented its restrictions within weeks of this change, a timeline that hasn’t gone unnoticed. Steinberger and investor Dave Morin attempted to negotiate an alternative solution with Anthropic, but according to them, they only managed to delay the enforcement by a single week.

      “First they copy popular features into their closed framework, then they lock out open-source,” Steinberger commented regarding the ban.

      Regardless of whether the timing is a strategic decision or mere coincidence, the result is the same. The most popular open-source agent framework, now loosely tied to OpenAI, has been effectively priced out of Claude’s subscription tier.

      The financial impact on users

      For developers who are used to unlimited agentic operations under a flat plan, the new billing system represents a considerable disruption. Under the pay-as-you