What's New in Claude · Released 28 September 2026
The everyday model got faster, and cheaper per task.
Claude Sonnet 5.5 is the second model in Anthropic’s 5.5 family, released on 28 September 2026. Anthropic says it runs more than 30% faster than Sonnet 5 and costs up to 30% less per task, because it needs fewer tokens to do the same work. Its list price is unchanged.
Anthropic positions it for well-scoped everyday work: polished documents, slides, spreadsheets and bug fixes. Opus 5.5 stays the model for complex, open-ended work that needs sustained judgment. Two days later Anthropic set a retirement date for Sonnet 4.5 on the Claude API: 30 November 2026.
What changed
claude sonnet 5.5 what's new
What Anthropic shipped.
Each figure is Anthropic’s, linked in the sources below.
- Speed and cost per task
- More than 30% faster output than Sonnet 5, and up to 30% less per task in Anthropic’s testing. On several benchmarks Sonnet 5.5 at low or medium effort beats Sonnet 5’s best score for about a tenth of the cost per task.
- Price
- $2 per million input tokens and $10 per million output tokens, the same as Sonnet 5. Cache reads were $0.20 per million at launch and were halved to $0.10 on 7 October.
- Where it sits
- Two points below Opus 5.5 on GDPval-AA, a test of real-world work across occupations. Anthropic is explicit that Opus 5.5 remains clearly stronger at complex, open-ended work.
- Sonnet 4.5 retirement
- Announced 30 September: claude-sonnet-4-5-20250929 retires from the Claude API on 30 November 2026. Anthropic recommends moving to Sonnet 5.5.
What it means
should my company use claude sonnet or opus
What it means for your company.
The question we hear most after a release is which model people should use. With Sonnet 5.5 the answer is easier to give to a whole team: Sonnet for the work you can describe in a sentence, Opus for the work you would hand to your most senior analyst.
- Leadership
- The cost of everyday AI work keeps falling while the list price stays flat. That makes it cheaper to put Claude in front of more people, which is where the return actually comes from.
- Your teams
- Board packs, client updates, reformatting a model output into a slide: Sonnet 5.5 is built for that work and is fast enough to iterate on. Keep Opus 5.5 for reading a long agreement or forming a view.
- IT and engineering
- Code written for Sonnet 5 can break in five ways, including forced tool use returning a 400 error and a new way to turn off up-front thinking. Anything still on Sonnet 4.5 through the API has until 30 November to move.
This week
What to do this week.
Three things, in this order.
- Find anything on Sonnet 4.5
- Ask whoever owns your integrations for every API call that names claude-sonnet-4-5. Each one needs a tested replacement before 30 November 2026.
- Write the one-line model rule
- Give your people a sentence they can remember: Sonnet for drafting and formatting, Opus for judgment. A rule nobody has to look up is a rule people follow.
- Re-run one real task
- Take a document task your team did last week and run it on Sonnet 5.5. Compare time and quality with what they produced. That comparison is worth more than any benchmark.
Sources
Where every figure on this page comes from.
Facts, prices and dates are Anthropic’s own, from the pages below. What it means and what to do are our view, from the firms we work with. Written by By The People Technologies and reviewed by Fatjon (Tony) Kalemaj, COO and Head of Applied AI, before it was published on 13 October 2026.
- Anthropic: Introducing Claude Sonnet 5.5
- www.anthropic.com
- Claude Platform release notes, 28 and 30 September 2026
- platform.claude.com
- Claude model deprecations
- platform.claude.com
- Claude models overview and pricing
- platform.claude.com
Questions
The short answers.
The questions people ask about this release, answered in full.
How much does Claude Sonnet 5.5 cost?
$2 per million input tokens and $10 per million output tokens on the Claude API, the same as Sonnet 5. Cache reads cost $0.10 per million tokens since 7 October 2026.
Is Claude Sonnet 5.5 better than Opus 5.5?
No. Anthropic places it two points below Opus 5.5 on GDPval-AA and says Opus 5.5 remains clearly stronger at complex, open-ended work. Sonnet 5.5 is faster and cheaper per task, which makes it the better default for well-scoped everyday work.
When is Claude Sonnet 4.5 retired?
Anthropic announced on 30 September 2026 that Claude Sonnet 4.5 (claude-sonnet-4-5-20250929) retires from the Claude API on 30 November 2026, and recommends migrating to Claude Sonnet 5.5.
Want this working inside your firm?
We run Claude workshops on your own documents and stay on as your teams adopt each release. Write to us and say what you are trying to get done.
Subject line: Claude: Claude Sonnet 5.5, and the Sonnet 4.5 retirement date