Image: Anthropic

Anthropic has opened up fast mode for Claude Opus 5.5 to most paying developers on its platform. The feature is no longer a "research preview," and the waitlist is gone, so developers no longer need to ask for access first.

Fast mode is exactly what it sounds like. Anthropic says it is the same Opus 5.5 model, with the same intelligence and behaviour, running on a faster setup that writes its answers up to 2.5 times quicker. You pay more per word for that speed.

Opus 5.5 fast mode in the Claude Platform is no longer in research preview, and we’ve removed the waitlist!

It’s the same Opus 5.5 but up to 2.5x faster, at $8 / $40 per million tokens.

It's also available in Microsoft Foundry (Hosted on Azure), with Google Cloud coming soon. https://t.co/oB0fhEgC88

— ClaudeDevs (@ClaudeDevs) October 15, 2026

The price is $8 per million input tokens and $40 per million output tokens, according to Anthropic's fast mode documentation. A token is a small chunk of text, roughly three-quarters of a word. That is exactly double the standard Opus 5.5 rate of $4 and $20 listed on Anthropic's pricing page.

There is a catch for anyone hoping for instant replies. Anthropic says the speed-up applies to how fast the answer streams out once it starts, not to the wait before the first word appears. So it helps most with long answers, such as big chunks of code.

The bigger story is how long this took. Anthropic first offered fast mode in February, on Opus 4.6, and it stayed behind a waitlist through Opus 4.7, Opus 4.8, Opus 5 and the first weeks of Opus 5.5, according to its release notes. Opus 5.5 is the first version anyone with a regular paid API account can switch on without approval.

It is also cheaper than before. Fast mode on Opus 5 and Opus 4.8 still costs $10 and $50 and is still waitlist-only. Fast Opus 5.5, at $8 and $40, now costs less than standard Claude Fable 5.1, Anthropic's pricier top-tier model, which lists at $10 and $50.

"Open to all" has limits, though. The documentation says fast mode has its own capacity limits, separate from normal Opus limits. When they're hit, requests are turned away for a short time, so access doesn't guarantee speed at busy moments. Organizations still in Anthropic's entry-level Evaluation tier have to move up a tier first.

Fast mode also isn't everywhere. Anthropic says it works on its own Claude API and on Microsoft Foundry deployments hosted on Azure, with Google Cloud "coming soon." The company hasn't given a date for Google Cloud. It doesn't work with the cheaper Batch API, and it isn't available on Claude Platform on AWS.

How to use it

Developers add "speed": "fast" to an Opus 5.5 request. The old beta header is no longer needed, though Anthropic says requests that still send it keep working. Each reply reports whether it actually ran fast or at standard speed.

In Claude Code, typing /fast turns it on, according to the Claude Code docs. Pro, Max, Team and Enterprise subscribers pay for it from usage credits rather than their plan's included usage, and Team and Enterprise owners must enable it first.

One practical tip from Anthropic: turn fast mode on at the start of a session. Switching speeds midway means the model can't reuse its saved work from earlier in the conversation, which costs more.

For most people chatting with Claude in the app, nothing changes. This is a paid speed option aimed at developers, who now have to weigh whether a faster answer is worth double the price.