"Key Takeaways"
- Anthropic released Claude Sonnet 5.5 on September 28, 2026. It is the second model in the Claude 5.5 family, after Opus 5.5.
- Anthropic says it runs more than 30% faster than Sonnet 5 and costs up to 30% less per task, mainly because it needs fewer tokens to do the same work.
- API prices per token are unchanged from Sonnet 5: $2 per million input tokens, $10 per million output tokens, and $0.20 per million for cache reads.
- Anthropic positions it for well-scoped everyday work, such as fixing bugs and producing polished documents, slides, and spreadsheets. It says Opus 5.5 is still clearly stronger for complex, open-ended work.
Summarize this article with
Claude Sonnet 5.5 is not Anthropic's smartest model. It is the one built to leave on all day.
Anthropic released it on September 28, 2026, six days after Claude Opus 5.5. The pitch is simple: strong everyday results, faster, and for less money per task. Here is what actually changed, what Anthropic is claiming, and where you should still reach for something bigger.
Claude Sonnet 5.5 in plain English
Anthropic's Claude family comes in three sizes. Opus is the heavyweight for hard, open-ended problems. Haiku is the small, low-cost option for high-volume work, and Anthropic says Haiku 5.5 will join the Claude 5.5 family in the coming weeks. Sonnet sits in the middle.
Anthropic describes Sonnet 5.5 as strongest at well-scoped everyday tasks, fixing bugs, and creating polished documents, slides, and spreadsheets. It also says the model has 'a sharp eye for design.'
Well-scoped is the key phrase. It means a job with a clear goal and clear edges: fix this failing test, turn these notes into a deck, clean up this spreadsheet. Not: work out our product strategy.
What changed from Sonnet 5
- Speed. Anthropic says Sonnet 5.5 generates output more than 30% faster than Sonnet 5, which makes it the fastest Sonnet model so far.
- Cost per task. The price per token has not changed. Anthropic says the model needs far fewer tokens to finish the same work, so a typical task costs up to 30% less.
- Writing. Anthropic says it writes more clearly than its previous generation of models, and that early testers described it as a better collaborator than Sonnet 5.
- Coding. Anthropic reports large gains on agentic coding tests. Early testers quoted in the announcement said it understood codebases quickly and finished tasks in fewer steps.
- Safeguards. Anthropic says it is the first Sonnet model to launch with the cybersecurity safeguards it uses on its most capable models. Higher-risk security tasks will visibly fall back to Sonnet 5. Routine bug fixing is unaffected, according to Anthropic.
Claude Sonnet 5.5 pricing, from Anthropic
| Price per 1M tokens | Claude Sonnet 5.5 | Claude Opus 5.5 |
|---|---|---|
| Input tokens | $2 | $4 |
| Output tokens | $10 | $20 |
| Cache reads | $0.20 | $0.20 |
| Cache writes | $2.50 | $5 |
What the pricing actually means for you
If you use Claude through the API, the list price is the same as Sonnet 5. The saving Anthropic describes comes from the model doing the same job in fewer tokens, not from a price cut. Your real saving depends on your tasks, so measure it on your own work before you rewrite a budget.
If you use Claude through the apps, you will mostly feel this as speed. Anthropic says the default effort setting is Medium in the Claude apps and Claude Code, and High on its developer platform. Lower effort answers faster and uses fewer tokens. Higher effort reasons for longer and checks its work more thoroughly.
How to read the benchmark numbers
Anthropic's headline figure is Terminal-Bench 4.0, an agentic coding test that measures multi-step work in a command-line environment. Anthropic reports 70.6% for Sonnet 5.5 against 10.3% for Sonnet 5.
That is a big jump. It suggests the model got much better at long, step-by-step terminal work. It does not mean Sonnet 5.5 is better at everything.
Anthropic says so itself. It writes that benchmark scores capture only one facet of a model, and that in its own and external testing Opus 5.5 'remains clearly stronger at complex, open-ended work requiring sustained judgment.'
Selected scores, as reported by Anthropic
| Test | Sonnet 5.5 | Sonnet 5 | Opus 5.5 |
|---|---|---|---|
| Terminal-Bench 4.0 (agentic coding) | 70.6% | 10.3% | 66.4% (highest effort) |
| CursorBench 4.0 (real coding sessions) | 55.5% | 34.1% | 57.8% |
| OSWorld 2.1 (computer use) | 80.1% | 57.0% | 81.8% |
What these numbers do not prove
They are vendor-reported, on tests the vendor chose, at specific effort settings. Anthropic's own footnotes note that Sonnet 5.5 scored lower at its Max effort setting than at a lower setting on one coding test.
Treat the chart as a reason to test, not a reason to switch. Your code, your documents, and your deadlines are still the real exam.
Where Claude Sonnet 5.5 fits in a normal week
- Fixing a bug that comes with a clear error message or a failing test.
- Turning meeting notes or a report into a first-draft deck using your own slide template.
- Cleaning, restructuring, or summarising a spreadsheet.
- Quick back-and-forth edits where waiting on a slower model breaks your flow.
Anthropic gives one internal example. It says it gave the model a public company's earnings materials, call transcripts, and a slide template, asked for a 10-slide operating review, and two experts judged the first draft ready to send. Treat that as a claim worth testing on your own material, not a promise.
For long, ambiguous projects, architecture decisions, or work where one bad judgment is expensive, Anthropic's own framing points to Opus 5.5. It describes Sonnet 5.5 as a complement to Opus, not a replacement.
How to try it this week
- Pick three real tasks you did last week with Sonnet 5 or another model.
- Run them again on Sonnet 5.5 at the default effort. Note how long each took and whether you had to fix the result.
- If you use the API, compare cost per task, not price per token. That is where Anthropic says the saving shows up.
- Developers: Anthropic lists the model ID as claude-sonnet-5-5. If you run Sonnet with thinking off, Anthropic says you need to switch to the new between_tools setting before moving. Check its migration guide first.
- Keep Opus for the task that genuinely needs deeper judgment.
Frequently Asked Questions
What is Claude Sonnet 5.5?
It is Anthropic's mid-sized model in the Claude 5.5 family, released on September 28, 2026. Anthropic says it is faster and cheaper per task than Sonnet 5, and best suited to well-scoped everyday work.
How much does Claude Sonnet 5.5 cost?
On the API, Anthropic lists $2 per million input tokens, $10 per million output tokens, $0.20 per million for cache reads, and $2.50 per million for cache writes. Anthropic says input, output, and cache-read prices match Sonnet 5, and that tasks cost less because the model uses fewer tokens.
Is Claude Sonnet 5.5 better than Claude Opus 5.5?
Not overall. Anthropic's reported scores show Sonnet 5.5 close to Opus 5.5 on several tests, and ahead on Terminal-Bench 4.0. But Anthropic says Opus 5.5 remains clearly stronger for complex, open-ended work. Sonnet 5.5 is the faster, cheaper choice for well-defined tasks.
Where can I use Claude Sonnet 5.5?
Anthropic says it is available on all its platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure. Developers can call it with the model ID claude-sonnet-5-5.
When is Claude Haiku 5.5 coming?
Anthropic says Haiku 5.5 will join the Claude 5.5 family in the coming weeks. It has not given a date.
The Chai Takeaway
Claude Sonnet 5.5 is the kind of upgrade you notice in your day, not on a leaderboard. Give it your ordinary, well-defined work, time it, and check the bill per task. Save Opus for the jobs that genuinely need it.
Explore More AI Tools