Claude Opus 5.5 is out and now the default model: what it changes for you

On September 22, 2026, Claude updated its top model: Opus 5.5 shipped, and it became the default. Let's unpack what this changes for you if you use Claude for work rather than for writing code.

What exactly shipped?

The Claude Code changelog (the official list of updates, code.claude.com/docs/en/changelog) records in version 2.1.280, dated September 22, that Claude Opus 5.5 was added with the identifier claude-opus-5-5, and it is now the default Opus model. In plain words, the top Claude line got a new version, and that is the one turned on by default.

The same entry notes separately that on the Pro and Team Standard plans the default model changed from Sonnet to Opus, matching what Max, Team Premium, and Enterprise already had. If you are on Pro, your default working model is now the top one, not the mid one.

What does a million tokens of context mean?

Opus 5.5 has a one-million-token context window. A token is a small chunk of text, roughly three quarters of a word in English. The context window is how much text the model keeps in mind during one conversation: your question, attached files, and all of its earlier answers in that chat.

A million tokens is on the order of a thousand-plus pages. In practice that means you can drop a fat contract, an annual report, and a thread of correspondence into a single chat, and Claude will answer while holding all of it at once rather than in scraps. Before, long documents had to be cut into parts and the answers stitched together by hand. Now you often do not need to cut.

Do I need to switch anything?

No. The default model is the one selected automatically. If you never changed the model in settings on purpose, Opus 5.5 is already running for you. Checking is simple: open the model switcher above the input field and see what is set right now.

If you used to pick Sonnet by hand for speed, you still can. Opus is stronger and a touch slower when it reasons; Sonnet is faster on simple tasks. The choice stays yours, the update just raised the bar for what sits there by default.

Did the price change?

If you are on Pro or Max, you pay a fixed monthly fee and do not pay extra per model. The default got stronger while your monthly payment did not move.

The prices in the changelog apply to the API (the interface for developers, access to the model through code with pay-per-volume). For the API, Opus 5.5 costs 4 dollars per million input tokens and 20 dollars per million output tokens, with cache reads at 0.20 dollars per million. If you do not call Claude through code, these numbers do not concern you at all.

Should you move to Opus 5.5?

You do not need to move anywhere: you are already on it if you did not touch the settings by hand. The practical point of the update is two things. First, Pro subscribers got the top model by default, with no surcharge and no extra steps. Second, the million-token window makes working with big documents easier, and that is half of office routine.

Everything else works as before: parsing contracts, consolidating reports, letters, proposals. There is simply a fresher model under the hood now, and on long files it loses the thread less often.

Want to learn how to squeeze the most out of Claude for your own tasks, start with our free materials: guides and the first lesson with no signup.

AGINE Academy is an independent product, not affiliated with Anthropic. Claude is owned by Anthropic.