Anthropic released Claude Sonnet 5.5 on September 28, 2026, just six days after launching Opus 5.5. This is the mid-tier model most people actually use day to day, and the update focuses on something simple: doing the same work faster and cheaper, without raising the price.
Quick Answer
- Claude Sonnet 5.5 released September 28, 2026, the second model in the Claude 5.5 family
- Pricing stays the same as Sonnet 5: $2 per million input tokens, $10 per million output tokens
- It runs more than 30% faster and can cost up to 30% less per task
- It is the first Sonnet model to launch with cybersecurity safeguards similar to Anthropic’s most capable models
- Available now through the Claude API, AWS, Google Cloud, Microsoft Azure, and Claude Code
- Claude Haiku 5.5, the smallest and cheapest model in the family, is expected in the coming weeks
What’s Actually New in Sonnet 5.5
Sonnet has always been Anthropic’s middle option: not as powerful as Opus, but stronger and more capable than Haiku, built for everyday tasks that still need real reasoning. Sonnet 5.5 keeps that role but performs noticeably better in two areas.
Speed: Output generation is more than 30% faster than Sonnet 5, making it the fastest Sonnet model Anthropic has released.
Cost per task: Even though the per-token price hasn’t changed, Anthropic says real-world tasks can cost up to 30% less overall. This comes from the model working faster and needing fewer back-and-forth tool calls to finish a job, not from a price cut itself.
Anthropic also says the model writes with clearer style, something early testers reportedly noticed as an improvement over the previous generation, particularly on less complex, fast-iteration tasks.
Pricing: Nothing Changed, and That’s the Point
| Claude Sonnet 5.5 | |
|---|---|
| Input tokens | $2 per million |
| Output tokens | $10 per million |
| Cached reads | $0.20 per million |
This is identical to Sonnet 5’s pricing. In an industry where prices have been moving constantly (Anthropic and OpenAI both cut prices on new model releases just last week, which we covered in our Claude Opus 5.5 vs GPT-6 comparison), keeping the price flat while improving speed is effectively a price cut on a per-task basis.
New Security Safeguards
Sonnet 5.5 is the first Sonnet model to ship with cybersecurity protections and fallback safety measures similar to those used on Anthropic’s more advanced models, closer to what Opus 5 has had. Anthropic says these protections, along with the model’s biology-related safeguards, which remain unchanged from Sonnet 5, are narrowly targeted at a small set of high-risk requests. Ordinary software development work and most life sciences use cases are not affected by these restrictions.
How It Performs
Anthropic shared a few benchmark comparisons:
- On AA-Briefcase, an agentic task benchmark, Sonnet 5.5 scores 1811, close behind Opus 5.5’s 1822
- On OSWorld 2.1, a computer-use test, Sonnet 5.5 reaches 80.1%, a sharp jump from Sonnet 5’s 57%, and close to Opus 5.5’s 81.8%
Anthropic also noted Sonnet 5.5 is the first Sonnet model able to beat the video game Pokémon Red using only screenshots as input, a test the company uses to check long, multi-step reasoning combined with visual understanding.
Should You Switch From Sonnet 5
If you already use Sonnet through the API, Claude Code, or one of the major cloud platforms, there is little reason not to move to 5.5. The price is unchanged, the performance is meaningfully better, particularly on tasks involving screen or computer interaction, and it is already rolling out as the default in tools like Claude Code.
Where to Access It
Claude Sonnet 5.5 is live now through:
- Anthropic’s own API
- Amazon Web Services
- Google Cloud
- Microsoft Azure
- Claude Code, where it is already available
What’s Coming Next
Anthropic confirmed that Claude Haiku 5.5, the smallest, fastest, and cheapest model in the Claude 5.5 lineup, built for high-volume and cost-sensitive work, is expected in the coming weeks. That will complete the full 5.5 family: Opus, Sonnet, and Haiku.
Frequently Asked Questions
When was Claude Sonnet 5.5 released?
September 28, 2026, six days after Claude Opus 5.5.
Is Claude Sonnet 5.5 more expensive than Sonnet 5?
No. Pricing is unchanged at $2 per million input tokens and $10 per million output tokens. Real-world costs can actually be lower due to faster performance.
How much faster is Sonnet 5.5 than Sonnet 5?
Anthropic says output generation is more than 30% faster, with real tasks costing up to 30% less overall.
What’s different about Sonnet 5.5’s safety features?
It is the first Sonnet model to include cybersecurity safeguards similar to Anthropic’s more capable models, targeted narrowly at high-risk requests without affecting normal development work.
Where can I use Claude Sonnet 5.5?
Through Anthropic’s API, AWS, Google Cloud, Microsoft Azure, and Claude Code.
Is Claude Haiku 5.5 out yet?
Not yet. Anthropic says it is expected in the coming weeks, completing the full Claude 5.5 model family.
How does Sonnet 5.5 compare to Opus 5.5?
Opus 5.5 remains the stronger model for the hardest tasks, while Sonnet 5.5 is the faster, lower-cost option for everyday work, with performance now much closer to Opus 5.5 than the previous Sonnet generation was.
