Anthropic has restricted unlimited access to its Claude AI models through third-party agents like OpenClaw, requiring heavy users to pay extra via API keys or usage bundles starting April 4, 2026. The policy shift, announced over the weekend, addresses severe system strain from high-volume agent tools previously covered under $20 monthly subscriptions.
On April 4 at 3 p.m. ET, Anthropic ended free usage of Claude subscriptions through third-party tools, as announced by Boris Cherny, head of Claude Code, on X. Heavy usage now incurs separate billing beyond the standard subscription, with options for discounted usage bundles or direct API access. AI product manager Aakash Gupta described it on X as the end of the 'all-you-can-eat buffet.'
Anthropic cited engineering constraints and outsized demand from agent tools like OpenClaw and Manis, which consume tokens rapidly in automated workflows—unlike Claude's primary design for human prompts. An email to customers stated: 'We've been working to manage demand across the board, but these tools put an outsized strain on our systems.' Prior measures included a five-hour session cap during peaks.
The company recently added features to Claude itself, such as remote computer access, echoing popular agent capabilities. Affected users can switch to alternatives like ChatGPT, Google Gemini, xAI, Perplexity, or DeepSeek, or use Anthropic's own Claude Cowork for workflows.
Criticism emerged from Peter Steinberger, OpenClaw's creator (now at OpenAI), who posted on X: 'Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source.' Cherny emphasized prioritizing direct customers and API users amid capacity challenges: 'Subscriptions weren't built for the usage patterns of these third-party tools.'
The move underscores tensions between AI labs managing costs and developers seeking open access to powerful models for autonomous agents.
