How to Use Claude Sonnet 5.5: Effort Settings, Cost and Limits

The setting that changes how Sonnet 5.5 feels is effort. Anthropic sets it to Medium by default in Claude Code and the Claude apps, but High on the Claude Platform. So the same model can behave quite differently depending on where you use it.

Everything here comes from Anthropic's own Sonnet 5.5 announcement, which I read in October 2026.

What Sonnet 5.5 is good at

• Everyday business work: documents, slides and spreadsheets that need to look finished.
• Coding fixes: Anthropic says the jump in performance is most noticeable in coding.
• Design: Anthropic says it has a sharp eye for design.
• Fast iteration: it's the fastest Sonnet model yet, which suits quick back-and-forth on simpler jobs.
• Long jobs and images: strong on long-horizon work and image understanding.

Anthropic is also clear about the limit: Opus 5.5 remains clearly stronger at complex, open-ended work that needs sustained judgment. That's the point to switch to Opus.

How to use Claude Sonnet 5.5

1. Open Claude on the web, desktop or mobile.
2. Choose Sonnet in the model menu.
3. Leave effort on Medium for routine work, which is the default in the apps.
4. Raise effort for harder tasks, when you want Claude to reason for longer and check its work.
5. Drop to Low for quick, simple jobs where speed and fewer tokens matter most.

Effort levels explained

Where you use it	Default effort	What that means
Claude apps	Medium	A balance of speed and quality for everyday chats
Claude Code	Medium	Faster coding sessions that use fewer tokens
Claude Platform (API)	High	Longer reasoning and more checking by default

Anthropic says lower settings answer faster and use fewer tokens, which suits routine work, while higher settings reason for longer and check their work more thoroughly. On several of its tests, Sonnet 5.5 at Low or Medium effort beat Sonnet 5's best score at a fraction of the cost per task.

What Claude Sonnet 5.5 costs on the API

Item	Sonnet 5.5, per million tokens
Input tokens	$2
Output tokens	$10
Cache reads	$0.10

Those are Anthropic's figures in US dollars, as of October 2026. Input and output prices are the same as Sonnet 5, but cache reads are half of Sonnet 5's $0.20, and Anthropic says Sonnet 5.5 usually needs far fewer tokens to do the same work, so a typical task costs less overall.

Before you migrate: the between_tools setting

Here's the developer detail that's easy to miss. Anthropic says that if you run Sonnet with thinking off, you'll need to switch to the new between_tools setting before moving to Sonnet 5.5. That setting keeps up-front thinking off. Read Anthropic's migration guide before you change the model name in production.

There's a second change for anyone who moves conversations between accounts. Sonnet 5.5 is the first Sonnet model to launch with safety classifiers that prevent reasoning extraction, and it expands preserved thinking, so Claude's thinking can't be separated from the account that created it. Anthropic says most developers won't notice, but switching accounts mid-session in Claude Code is affected.

How close Sonnet 5.5 gets to Opus 5.5

On GDPval-AA, a test of real-world work across many occupations, Anthropic says Sonnet 5.5 scores two points below Opus 5.5. On CursorBench, which uses tasks from real coding sessions, its best score is within about two points of Opus 5.5. At High effort on FrontierCode it scores 10 points higher than Sonnet 5 at the same setting, for about one fifteenth of the cost per task.

Put simply, for a lot of everyday work the gap to Opus is small, and the saving is real. That's why I'd start on Sonnet and only move up when a job keeps going wrong.

Security work may fall back to Sonnet 5

Because Sonnet 5.5's cyber abilities are a big step up, Anthropic has given it safeguards similar to Opus 5.5's. You can still find and fix bugs as part of normal software work, but higher-risk cybersecurity tasks visibly fall back to Sonnet 5. Its biology safeguards are the same as Sonnet 5's, and Anthropic says some microbiology and virology requests may be flagged in error.

Sonnet 5.5 for writing and content

Anthropic says Sonnet 5.5 writes more clearly than the previous generation, and early testers called it a better collaborator than Sonnet 5. For blog drafts, client emails and social posts, it's the model I'd try first before paying the extra for Opus. My guide to the best Claude model for writing goes deeper, and Claude Opus vs Sonnet compares the two families.

Sonnet 5.5 is available on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure, and it's available with zero data retention.

My verdict

Make Sonnet 5.5 your default for everyday work. It's fast, it writes clearly and it's cheaper per task than the model it replaces. Keep effort on Medium in the apps, raise it when the job is hard, and step up to Opus 5.5 for open-ended problems. If you build on the API, switch to between_tools first if you've been running with thinking off.

Frequently asked questions

When was Claude Sonnet 5.5 released?

Anthropic announced Claude Sonnet 5.5 on 28 September 2026 as the second model in its Claude 5.5 family, after Opus 5.5.

What is the default effort for Claude Sonnet 5.5?

Medium in Claude Code and the Claude apps, and High on the Claude Platform. You can raise or lower it depending on the task.

How much does Claude Sonnet 5.5 cost on the API?

Anthropic lists $2 per million input tokens and $10 per million output tokens, with cache reads at $0.10 per million, in US dollars as of October 2026.

Is Sonnet 5.5 as good as Opus 5.5?

Not for everything. Anthropic says Sonnet 5.5 at Max effort is comparable to Opus 5.5 on several tests, but Opus 5.5 remains clearly stronger at complex, open-ended work.

Do I need to change anything to switch to Sonnet 5.5?

If you run Sonnet with thinking off, Anthropic says you'll need to switch to the new between_tools setting first. Check Anthropic's migration guide before changing the model name.