- Claude Sonnet 5.5 came out on 28 September 2026. It is Anthropic’s new mid-range model and the second member of the Claude 5.5 family, after Claude Opus 5.5.
- Anthropic says it is 30%+ faster than Sonnet 5 and costs up to 30% less per task, with big jumps on coding, office work and computer-use tests — on several of them it lands close to Opus 5.5.
- Free Claude users can use Sonnet models, and reports say Sonnet 5.5 is on the free plan, unlike Opus 5.5.
- Developer price stays at $2 / $10 per million input/output tokens — half the price of Opus 5.5. Claude Haiku 5.5 follows “in the coming weeks”.
Anthropic has released Claude Sonnet 5.5, six days after Claude Opus 5.5. Sonnet is the model most Claude users actually talk to every day, so this is the Claude 5.5 upgrade that reaches the most people. Here is what changed, who can use it and how it compares with Opus 5.5, without the jargon.
What is Claude Sonnet 5.5?
Anthropic makes Claude models in several sizes. Opus is the careful, more expensive option for complex work; Sonnet is the faster, cheaper everyday model; Haiku is the smallest and cheapest. In its announcement, Anthropic calls Sonnet 5.5 “a clear upgrade over Claude Sonnet 5” and a “faster, lower-cost complement to Claude Opus 5.5”.
Anthropic says Sonnet 5.5 is strongest at:
- well-defined everyday tasks,
- fixing bugs in code,
- creating polished documents, slides and spreadsheets — Anthropic says it has “a sharp eye for design”.
Opus 5.5 remains the better choice, in Anthropic’s words, for “complex, open-ended work requiring sustained judgment”.
What’s new compared with Sonnet 5
Faster and cheaper to run
Anthropic says Sonnet 5.5 writes its answers more than 30% faster than Sonnet 5, making it the fastest Sonnet model so far. It also needs fewer tokens to finish the same job, which Anthropic says makes it up to 30% cheaper per task even though the price per token has not changed. Several companies that tried it early report similar effects: Slack says it did better on its Slackbot tests with about 14% fewer output tokens, and Zendesk says support tickets were processed 20% faster.
Much better at coding and computer tasks
The biggest jumps are in agentic work — where the AI runs commands, uses apps or edits code on its own. Anthropic’s published figures:
| Test (as reported by Anthropic) | Sonnet 5.5 | Sonnet 5 | Opus 5.5 |
|---|---|---|---|
| Terminal-Bench 4.0 (coding in a terminal) | 70.6% | 10.3% | 66.4% |
| CursorBench 4.0 (real coding sessions) | 55.5% | 34.1% | 57.8% |
| FrontierCode 1.1 (Main) | 46.2% (Max effort) | 42.4% | 54.4% |
| OSWorld 2.1 (using a computer) | 80.1% | 57.0% | 81.8% |
| GDPval-AA v2.1 (office work, Elo score) | 1844 | 1449 | 1846 |
| Humanity’s Last Exam (with tools) | 64.5% | 54.9% | 67.7% |
| Chartography (reading charts) | 61.6% | 15.6% | 64.4% |
Two things stand out. First, on the office-work test GDPval-AA, Sonnet 5.5 is almost level with Opus 5.5 and about 400 points ahead of Sonnet 5. Second, on Terminal-Bench 4.0 it scores above Opus 5.5’s reported result. Anthropic still stresses that benchmarks capture “only one facet” of a model and that Opus 5.5 is clearly stronger at complex work. These are the company’s own numbers; independent results will follow.
One fun detail from the announcement: Sonnet 5.5 is the first Sonnet model to finish the video game Pokémon Red working only from screenshots.
Better writing and design
Anthropic says Sonnet 5.5 picks up some of Opus 5.5’s clearer writing style. Early testers described it as a better partner for back-and-forth work and noted it adds polish to user interfaces and follows slide templates well. In one of Anthropic’s own internal trials, two experts judged a 10-slide operating review it drafted from a company’s earnings materials ready to send as is.
Is Claude Sonnet 5.5 free?
Claude’s pricing page lists Sonnet and Haiku models for the Free plan, while Opus is limited to paid plans. Notebookcheck reports that free Claude users get Sonnet 5.5 on the web and in the apps, unlike Opus 5.5. In practice:
| Claude plan | Sonnet 5.5 | Opus 5.5 |
|---|---|---|
| Free | Yes (with the free plan’s usage limits) | No |
| Pro, Max | Yes | Yes |
| Team, Enterprise | Yes | Yes |
Anthropic says the Claude apps and Claude Code run Sonnet 5.5 at Medium effort by default, a setting that balances speed and quality. Paid plans simply let you send more messages before hitting the limit. If you are deciding between Claude and ChatGPT, our Claude vs ChatGPT comparison covers the differences beyond the model.
Sonnet 5.5 vs Opus 5.5: which should you pick?
| Claude Sonnet 5.5 | Claude Opus 5.5 | |
|---|---|---|
| Released | 28 September 2026 | 22 September 2026 |
| Best for | Everyday tasks, bug fixes, documents and slides | Complex, open-ended work needing judgment |
| Speed | 30%+ faster than Sonnet 5 | 30%+ faster than Opus 5 |
| Free plan | Yes | No |
| API price (input / output per 1M tokens) | $2 / $10 | $4 / $20 |
| Cache reads / writes per 1M tokens | $0.20 / $2.50 | $0.20 / $5 |
- Use Sonnet 5.5 for most things: emails, summaries, spreadsheets, slides, everyday coding. It is quick and, for free users, it is the model you get.
- Switch to Opus 5.5 (paid plans) for long, messy problems where you want the most careful answer, such as planning a large project or reviewing a complicated contract.
A game developer quoted by Anthropic summed up the split: let Opus 5.5 design the architecture, then let Sonnet 5.5 build it. Our earlier article covers what Claude Opus 5.5 changed for users.
What Claude Sonnet 5.5 costs developers
On the Claude Platform the model ID is claude-sonnet-5-5. Pricing is unchanged from Sonnet 5: $2 per million input tokens, $10 per million output tokens, $0.20 for cache reads and $2.50 for cache writes. It is also available on Amazon Web Services, Google Cloud and Microsoft Azure, with zero data retention available.
That $2 / $10 price is the same as OpenAI’s new GPT-6.1 Sol and Google’s introductory price for Gemini 4 Argon — see Gemini 4 Argon vs GPT-6.1 Sol vs Claude Sonnet 5.5. To estimate a monthly bill, use the AI API cost calculator, and compare specs across providers in the AI models database.
Two technical notes from Anthropic’s announcement for developers moving from Sonnet 5:
- If you run Sonnet with thinking turned off, you need to switch to the new
between_toolssetting first; Anthropic’s migration guide has details. - Sonnet 5.5 is the first Sonnet model with cyber safeguards like Opus 5.5’s: routine bug-fixing is unaffected, but higher-risk security requests visibly fall back to Sonnet 5.
What about Claude Haiku 5.5?
Anthropic says Claude Haiku 5.5, built for high-volume and cost-sensitive use, will join the Claude 5.5 family “in the coming weeks”. No price or date has been announced.
Frequently asked questions
When was Claude Sonnet 5.5 released?
On 28 September 2026. It is available in the Claude apps on web, iOS and Android, in Claude Code, on the Claude Platform, and on AWS, Google Cloud and Microsoft Azure.
Is Claude Sonnet 5.5 free?
Claude’s Free plan includes Sonnet models, and Notebookcheck reports that free users get Sonnet 5.5, with the free plan’s usage limits. Opus 5.5 needs a paid plan.
Is Claude Sonnet 5.5 better than Opus 5.5?
Not overall. On Anthropic’s tests it comes close to Opus 5.5 and even beats Opus 5.5’s reported score on Terminal-Bench 4.0, but Anthropic says Opus 5.5 remains clearly stronger at complex, open-ended work. Sonnet 5.5 is faster and half the price.
How much does the Claude Sonnet 5.5 API cost?
$2 per million input tokens and $10 per million output tokens, the same as Sonnet 5, with cache reads at $0.20 and cache writes at $2.50 per million tokens.
How is Sonnet 5.5 different from Sonnet 5?
Anthropic says it is 30%+ faster, up to 30% cheaper per task because it uses fewer tokens, much stronger on coding and computer-use tests, and writes more clearly.
When will Claude Haiku 5.5 come out?
Anthropic only says “in the coming weeks”.