Default Fast Mode Rate Limits Added for Claude Opus 5.5

This documentation page outlines throughput controls, speed configurations, and model access policies across the platform. The update refines access rules for accelerated response handling across different generations of the Opus model family. Previously, high-speed execution was classified generally as a research preview. Under the revised documentation, standard automated allocations for accelerated execution are provided automatically to Claude Opus 5.5 accounts across the Start, Build, and Scale service tiers without requiring special permission. Customers in the basic Evaluation tier cannot access fast processing for this model unless they upgrade to a qualifying tier. Meanwhile, fast processing for Opus 5 and Opus 4.8 continues to operate under research preview status, and older clarifying statements regarding versions 4.6 and 4.7 have been eliminated.

In practice, developers with eligible subscription tiers can employ accelerated generation on Claude Opus 5.5 immediately, and those calls draw from an independent allocation separate from ordinary generation limits. If an organization exhausts its throughput, the interface will generate rate-limit response codes accompanied by backoff timing instructions. What the documentation does not state directly is the specific numerical request or token allotment assigned to each tier, directing developers to inspect their administrative dashboards instead.