Anthropic Takes Claude Opus 5.5 Fast Mode Out of Research Preview, Dropping the Beta Header and Access Request, at $8 and $40 per Million Tokens

Anthropic says Claude Opus 5.5 fast mode no longer needs an access request or beta header, adds a default rate limit in three usage tiers, and extends it to Microsoft Foundry on Azure.

anthropic claude-opus-5-5 fast-mode claude-api api-pricing

Verified pipeline

Sources: 3 Publisher: signed Contributor: signed Hash: 398a47b717 View

Editor's Note · October 17, 2026

Clarification:

The article gives fast mode and standard prices for Claude Opus 5.5 ($8 and $40 versus $4 and $20 per million input and output tokens) without stating when they were retrieved. These figures reflect Anthropic's pricing page and fast mode documentation as read on October 17, 2026; the pricing page itself directs readers to claude.com/pricing for the most current prices, and the figures may change.

Overview

Anthropic has moved fast mode for Claude Opus 5.5 out of research preview. In its October 15, 2026 entry, the Claude Platform release notes state that “Fast mode is out of research preview for Claude Opus 5.5, and you no longer need to request access.” The same entry says requests to Claude Opus 5.5 no longer require the fast-mode-2026-02-01 beta header, and that requests which still send the header keep working.

What Changed

Access. According to the release notes, Claude Opus 5.5 now has a default fast mode rate limit in the Start, Build, and Scale usage tiers. The fast mode documentation adds that organizations in the Evaluation tier can use fast mode with Claude Opus 5.5 once they move to the Start, Build, or Scale tier.
Header. The documentation says fast mode for Claude Opus 5.5 does not need the fast-mode-2026-02-01 beta header or an access request. Developers opt in by setting speed: "fast" on the request.
Older models. Per the release notes, fast mode for Claude Opus 5 and Claude Opus 4.8 remains in research preview, and developers should keep sending the beta header with fast mode requests to those models.
Microsoft Foundry. The release notes also say fast mode for Claude Opus 5.5 is now available on Claude in Microsoft Foundry for deployments hosted on Azure. The documentation states that fast mode is not available for Claude in Microsoft Foundry deployments hosted on Anthropic.

What Fast Mode Does

The documentation says “Fast mode delivers up to 2.5x higher output tokens per second from Claude Opus 5.5, Claude Opus 5, and Claude Opus 4.8 at premium pricing.” It describes the feature this way: “Fast mode runs the same model with a faster inference configuration.” The page lists no change to intelligence or capabilities, and says the speed benefits are focused on output tokens per second rather than time to first token.

The response’s usage object includes a speed field showing whether a request ran at "fast" or "standard" speed, according to the documentation. Fast mode has a dedicated rate limit separate from standard Opus rate limits, and exceeding it returns a 429 error with a retry-after header, per the same page.

Pricing

The pricing page lists fast mode for Claude Opus 5.5 at $8 per million input tokens and $40 per million output tokens. The same page lists standard Claude Opus 5.5 pricing at $4 per million input tokens and $20 per million output tokens, so the fast mode rates are double the standard rates. Claude Opus 5 and Claude Opus 4.8 are listed at $10 and $50 in fast mode.

According to the pricing page, fast mode pricing applies across the full context window, including requests over 200k input tokens, and prompt caching multipliers and data residency multipliers apply on top of it. The page also says fast mode is available on the Claude API and, for Claude Opus 5.5, on Claude in Microsoft Foundry for deployments hosted on Azure, and that it is not available on Claude Platform on AWS or partner-operated cloud platforms.

What We Don’t Know

The release notes and documentation do not say what the default fast mode rate limits are for each usage tier; the documentation points developers to the Rate limits page in the Claude Console. The “up to 2.5x” figure is an upper bound, and the sources do not give measured typical speedups for Claude Opus 5.5 specifically.

The Machine Herald previously reported on the launch of Claude Opus 5.5 itself. A similar speed tier from another provider was covered in a separate article on OpenAI’s Ultrafast mode for GPT-6.1 Sol.