Opus 5.5 fast mode drops its access request, keeps its premium price

A long response is useful sooner only if you are actually waiting to use it. Anthropic’s October 15, 2026 API update removes the access request for Opus 5.5 fast mode and stops requiring its beta header. Existing requests with that header still work. This changes access to a speed option, rather than launching another model. Platform release notes.

Measure the wait that matters

Fast mode uses the same model weights and capabilities. Anthropic advertises up to 2.5 times higher output tokens per second; that is a vendor claim about generation speed, not first-token latency or the time needed to finish and check a whole job. How fast mode works.

For a proposed comparison, use one known coding task with the same context and acceptance checks. Record when the usable change is ready, how much review it needs and the request’s usage.speed. A quick stream of text that needs another repair pass may not improve the workday. We have not performed this comparison.

Price a representative request before changing the default

Current Opus 5.5 fast rates are $8 per million input tokens and $40 per million output tokens, versus $4 and $20 at standard speed. An illustrative uncached request with 10,000 input and 2,000 output tokens therefore costs $0.16 fast or $0.08 standard, before other modifiers, tools or retries. Prompt caching and data-residency modifiers can affect the bill. Current pricing.

Fast mode has separate capacity limits and can return a rate-limit response. The new release includes Azure-hosted Microsoft Foundry deployments, but the documentation excludes Claude Platform on AWS and does not make every cloud deployment eligible. Earlier Opus versions retain their own preview rules. Availability and limits.

Our Opus 5.5 launch coverage describes the underlying model, and Claude Code workflow guide focuses on completing verified changes. Keep standard speed where paying more does not remove a meaningful wait.