Claude Opus 5.5 is included with Pro, and Free still misses out

Anthropic released Claude Opus 5.5 on Tuesday, the first model in its 5.5 family. The announcement leads with speed and price. The part that matters to subscribers sits further down. Anthropic is raising the five-hour usage limits on Pro, Max, Team and seat-based Enterprise plans. Every subscriber also gets a rate limit reset that can be saved and spent whenever they want.

Who gets to see it

Anthropic's pricing page now lists Opus as a plain yes on Pro, Max and Team, and a no under Free. By that table a free account still gets Sonnet and Haiku. The line is blurrier in practice. In mid-September our colleague Yuval Goel found that Opus 5 had quietly turned up on free accounts, capped at three messages and outside the usual five-hour reset, with no announcement from Anthropic.

The previous release worked differently. Everyone on a paid plan could see Fable 5.1 in the model menu, but Pro subscribers could only run it through usage credits, billed on top of the subscription. Opus 5.5 comes with no such condition on Pro.

One more thing changed on that page. Opus 5 has dropped out of the current models and now sits under "Legacy models" alongside Fable 5 and the older 4-series releases. Anthropic gives no shutdown date for it. Sonnet 5.5 and Haiku 4.5 are due in the coming weeks.

What the new pricing looks like

Through the API, a million input tokens now costs $4 and a million output tokens $20, each 20% below Opus 5. Cache reads fall further, from $0.50 to $0.20 per million tokens, and those carry most of the cost in agentic and coding work. Anthropic puts the combined saving at roughly 40% on typical workloads, helped by the model spending fewer tokens per task. An optional fast mode runs up to 2.5 times faster and costs twice as much. None of this reaches your monthly bill as a subscriber, because a subscription meters limits rather than tokens. Those limits are what went up.

Two conditions the announcement glosses over

Thinking mode can no longer be switched off on Opus 5.5, which Anthropic states in its developer documentation. Safeguards also act during the conversation itself. Since the model matches Claude Mythos 5.1 in biology and cybersecurity, most cybersecurity tasks are routed to the older Opus 4.8 instead. Anthropic calls the fallback transparent. What it means for you is that some answers arrive from a weaker model with no indication in the chat window.

The model does well on one front that matters more than benchmark points. On a test run by the security firm Gray Swan, Opus 5.5 ties Fable 5.1 for the lowest prompt injection success rate of any model tested. That counts as soon as Claude reaches your mail, calendar or files through connectors, which is the default arrangement since Cowork and chat were folded into one app.

What to do with it

Open Usage in your settings. It shows where your current five-hour window stands and whether the reset has landed in your account. The announcement omits a detail worth knowing. In the app the reset expires on Oct. 22. Save it for one long session in the next four weeks and it earns its keep. Leave it alone and it is gone. Anyone who had pinned Opus 5 should check the model menu, because the old selection stays put and nothing switches over on its own.