Anthropic carves all non-interactive use out of monthly subscriptions (venturebeat.com)

Anthropic has announced a significant change to its Claude AI subscription model, reintroducing support for third-party autonomous agents like OpenClaw by launching a new "Agent SDK" credit system. This system allows subscribers to allocate a specific amount of credits per month (ranging from $20 to $200, depending on the subscription tier) for programmatic uses, such as powering external agents. This marks a reversal from the company's earlier policy, which banned such uses due to concerns over excessive demand on their computational resources. The new credits differentiate between "interactive" and "programmatic" usage, enforcing a strict cap on the amount of resources available for non-interactive tasks. If users exhaust their monthly credits, they must purchase additional usage at standard API rates. While this creates a more sustainable financial model for Anthropic, it also has drawn criticism from the developer community, who feel that the changes essentially devalue their previous subscriptions and limit usage. The shift reflects Anthropic’s efforts to balance user demands with operational constraints, although it has sparked frustration among developers who relied heavily on unrestricted access for automation tasks.