Home / Technology / Anthropic Restricts Claude Subscriptions for AI Agents
Anthropic Restricts Claude Subscriptions for AI Agents
4 Apr
Summary
- Claude Pro and Max subscriptions can no longer power third-party AI agents starting April 4, 2026.
- Users must now opt for pay-as-you-go 'extra usage' or the API for external agent connections.
- The change aims to manage compute strain and prioritize direct Anthropic product users.
Starting Saturday, April 4, 2026, at 12 pm PT, Anthropic will no longer permit Claude Pro and Max subscribers to use their plans for third-party AI agents. This policy change impacts users connecting services like OpenClaw to Claude models.
Anthropic cited significant strain on its compute and engineering resources as the primary reason for this decision. Boris Cherny, Head of Claude Code at Anthropic, explained that subscription models were not designed for the usage patterns of these external tools, necessitating a shift to ensure reliable service for all customers.
To continue using Claude models with external agents, users will need to adopt a pay-as-you-go 'extra usage' billing system or leverage Anthropic's API. These methods charge based on actual token usage rather than offering the unlimited-style access previously available through Pro and Max plans.
This adjustment follows recent stricter session limits imposed by Anthropic, which affected token allowances during peak hours to manage growing demand. The company asserts that these changes do not entirely ban third-party tools but rather move them to a different, usage-based financial model.
