Short Answer: What is Claude Sonnet 5.5?
Anthropic launched Claude Sonnet 5.5 on September 28, 2026, the second model in the Claude 5.5 family, calling it a clear upgrade over Sonnet 5 that runs more than 30% faster and costs up to 30% less for most work. It is priced the same as Sonnet 5 per token but typically needs far fewer tokens to finish a task, and it is positioned as the faster, lower-cost complement to Claude Opus 5.5.
Quick Summary
- Claude Sonnet 5.5 launched September 28, 2026, following Opus 5.5 by less than a week as the second model in the new Claude 5.5 family.
- Anthropic says it is a clear upgrade over Sonnet 5: 30%+ faster output generation and up to 30% lower cost per task for typical work, at the same per-token pricing.
- On Terminal-Bench 4.0, an agentic coding benchmark, Sonnet 5.5 scores 70.6% versus Sonnet 5’s 10.3% — and it beats Opus 5.5’s 66.4% on that specific test.
- It is the first Sonnet model to launch with cybersecurity safeguards and fallbacks similar to those built for Anthropic’s most capable models, since its cyber capabilities now match Opus 5’s.
- Claude Haiku 5.5, the third and final model in the 5.5 family, is expected within weeks.
Anthropic introduced Claude Sonnet 5.5 on September 28, 2026, positioning it as a faster, lower-cost complement to Opus 5.5 rather than a replacement for it. Where Opus 5.5 is built for complex work requiring careful judgment, Anthropic says Sonnet 5.5 is strongest at well-scoped everyday tasks: fixing bugs, and creating polished documents, slides, and spreadsheets. The company also says it has “a sharp eye for design.”
How Does Sonnet 5.5 Perform on Benchmarks?
Anthropic’s published results show Sonnet 5.5 jumping dramatically past Sonnet 5 on agentic coding tasks, and landing close to — in a few cases ahead of — Opus 5.5 on several evaluations:
| Sonnet 5.5 | Sonnet 5 | Opus 5.5 | GPT-6 Sol | |
|---|---|---|---|---|
| Agentic coding Terminal-Bench 4.0 |
70.6% | 10.3% | 66.4%¹ | — |
| Agentic coding FrontierCode 1.1 (Main) |
46.2% Max² 52.1% Xhigh |
42.4% | 54.4% | 49.3% |
| Agentic coding CursorBench 4.0 |
55.5% | 34.1% | 57.8% | — |
| Knowledge work GDPval-AA v2.1³ |
1844 | 1449 | 1846 | 1487⁴ |
| Knowledge work AA-Briefcase v1.1³ |
1811 | 1359 | 1822 | 1483⁴ |
| Multidisciplinary reasoning Humanity’s Last Exam |
64.5% with tools |
54.9% with tools |
67.7% with tools |
— |
| Computer use OSWorld 2.1 |
80.1% partial |
57.0% partial |
81.8% partial |
— |
| Visual chart recognition Chartography |
61.6% no tools |
15.6% no tools |
64.4% no tools |
53.6%⁴ no tools |
¹ Terminal-Bench 4.0 result for Opus 5.5 at Xhigh effort, its highest score. ² Sonnet 5.5 scores lower at Max effort than Xhigh on FrontierCode; Anthropic attributes this to the model more often invoking a multi-subagent code-review skill at Max effort, which occasionally caused timeouts or out-of-scope edits. ³ Figures reflect a pre-release Sonnet 5.5 deployment with a structured-output bug Anthropic says has since been fixed. ⁴ GPT-6 Sol scores may not yet reflect a recent OpenAI fix to an image-understanding bug. Source: Anthropic.
Anthropic is upfront that benchmark scores only capture one facet of capability, and that in its own testing and external testers’ testing, Opus 5.5 remains clearly stronger at complex, open-ended work requiring sustained judgment. The more consistent story across the results is efficiency: on several benchmarks, Sonnet 5.5 at Low or Medium effort beats Sonnet 5’s best score for roughly a tenth of the cost per task.
How Much Faster and Cheaper Is It?
| 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 |
Sonnet 5.5 is priced identically to Sonnet 5 per token, but Anthropic says it typically needs far fewer tokens to do the same work, which nets out to up to 30% lower cost per task in testing. It also generates output more than 30% faster than Sonnet 5, making it Anthropic’s fastest Sonnet model to date. Effort can still be adjusted to trade cost and speed against quality: Claude Code and the Claude apps default to Medium effort, while the Claude Platform defaults to High.
How Good Is It at Coding?
Anthropic’s coding-specific numbers are the sharpest jump in the release: at High effort on FrontierCode, Sonnet 5.5 scores 10 points higher than Sonnet 5 at the same setting, for about one-fifteenth the cost per task. On CursorBench, its best score lands within about two points of Opus 5.5.
Early testers echoed that in practice, not just on benchmarks:
- Epic Games: “Claude Sonnet 5.5 cleared the same quality bar you’d expect from a higher-tier model, holding up on a system design audit and a data flow review… managed tens of thousands of lines of code for gameplay system architecture, kept responses snappy, handled multi-hour tasks, and delivered with less prescriptive prompting.” — Daniel Vogel, COO
- CodeRabbit: “Sonnet 5’s tendency to reach for web search too often and its high token use are both gone in this new model. We plan to move simple and moderate reviews over now, and more in the coming weeks.” — David Loker, VP of AI
- Base44: “Across 118 real app builds, Claude Sonnet 5.5 produced apps that scored level with Opus 5. It got there in 3.6 iterations per build on average, where Opus 5 took 7.7.” — Gabriel Grinberg, AI Engineering Lead
- Unity: “It also completed 90% of tasks in our multi-step Unity Editor and coding benchmark, beating similar models.” — Sam Zhang, Creative Technologist
How Does It Handle Knowledge Work?
On GDPval-AA, which tests real-world tasks across 44 occupations and nine industries, Sonnet 5.5 scores nearly level with Opus 5.5 and roughly 400 points above Sonnet 5. Anthropic says early testers also called out less quantifiable gains: a more natural conversational style and a knack for design, including slide decks built from a template that needed minimal editing. In one internal test, two experts judged Sonnet 5.5’s first-draft 10-slide operating review, built from earnings materials and a template, ready to send as-is.
- Slack: “Without changing any of our prompts, Claude Sonnet 5.5 did better than Sonnet 5 on almost all of our offline Slackbot evals, in fewer steps and with about 14% fewer output tokens.” — Curtis Allen, Principal Engineer
- Zendesk: “Tickets were processed 20% faster, getting our customers the help they need without the wait.” — Abhinay Kathuria, Director of AI
- Balyasny Asset Management: “On our private suite of 2,441 finance tasks… Claude Sonnet 5.5 scored ahead of Sonnet 5 and used about 121k tokens per answer, where Sonnet 5 used 497k.” — Joe Poirier, Senior AI Engineer
- Box: “Compared to the last model, Sonnet 5.5 was more accurate, 2.4x faster, and used 12% fewer total tokens.” — Yashodha Bhavnani, VP of AI Products
What Changed for Safety?
Because Sonnet 5.5 does not push Anthropic’s overall capability frontier, the company says its alignment assessment focused on a narrower set of risks common to models at any capability level, such as acting against a user’s interests or cooperating with high-stakes misuse, rather than the more expansive review given to Opus 5.5. On Anthropic’s automated behavioral audit, run across roughly 1,850 scenarios, Sonnet 5.5 improves on or matches Sonnet 5 on most alignment, misuse-resistance, and honesty measures. On newer containment evaluations, it comes close to Opus 5.5 (the strongest model Anthropic tested) in how rarely it tries to escape its sandbox, and it’s the least likely of any Anthropic model to probe the limits of its containers.
Because its cybersecurity capabilities now roughly match Opus 5’s, Sonnet 5.5 is the first Sonnet model to launch with cyber safeguards and fallbacks similar to those built for Anthropic’s most capable models: higher-risk cybersecurity tasks are automatically routed to the older Sonnet 5, while routine bug-fixing in normal development work is unaffected. Its biology safeguards are unchanged from Sonnet 5’s. It also launches with safety classifiers that block reasoning-extraction attempts and expanded “preserved thinking” protections, making it the first Sonnet model to ship with anti-distillation defenses of that kind.
Where Can You Access It?
Claude Sonnet 5.5 is available now on all platforms, including the Claude Platform (as claude-sonnet-5-5), Amazon Web Services, Google Cloud, and Microsoft Azure, with zero data retention available as with prior models. It joins Opus 5.5, which launched September 22, as the second model in the Claude 5.5 family; Claude Haiku 5.5 is expected to complete the lineup within weeks. The new model also slots directly into the kind of long-running, multi-agent work Anthropic has been building toward with Claude Code Projects and the internal automation the company has disclosed at Anthropic itself.
Why This Matters
The headline efficiency numbers matter less on their own than what they signal about Anthropic’s release cadence: a second frontier-adjacent model shipped within a week of the first, both cheaper and faster than their predecessors, both improved on alignment measures rather than trading safety for speed. For most day-to-day agentic and knowledge work, Sonnet 5.5’s benchmark scores landing within a few points of Opus 5.5, at roughly half the price, makes it the more relevant model for most teams — Opus 5.5 remains the one to reach for when a task genuinely needs sustained, open-ended judgment rather than fast, well-scoped execution.
Frequently Asked Questions
How is Claude Sonnet 5.5 different from Claude Opus 5.5?
Sonnet 5.5 is positioned as the faster, cheaper model for well-scoped everyday tasks, bug fixes, and polished documents, while Opus 5.5 is built for complex, open-ended work requiring sustained judgment. On several benchmarks Sonnet 5.5 scores close to or even above Opus 5.5, but Anthropic says Opus 5.5 remains clearly stronger on the hardest, most open-ended tasks.
How much cheaper is Sonnet 5.5 than Sonnet 5?
Per-token pricing is unchanged ($2 per million input tokens, $10 per million output tokens), but Sonnet 5.5 typically needs far fewer tokens to complete the same task, which Anthropic says works out to up to 30% lower cost per task in testing, alongside 30%+ faster output generation.
Is Claude Sonnet 5.5 safer than Sonnet 5?
Anthropic says Sonnet 5.5 improves on or matches Sonnet 5 on most measures of alignment, misuse resistance, and honesty, and is the least likely of any Anthropic model to probe the limits of its sandbox in containment testing. Because its cyber capabilities are much stronger than Sonnet 5’s, it is also the first Sonnet model to launch with safeguards that route higher-risk cybersecurity tasks to an older model.
Where is Claude Sonnet 5.5 available?
It is available now on the Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure. Claude Haiku 5.5 is expected to join the Claude 5.5 family within weeks.
Does Sonnet 5.5 replace Opus 5.5?
No. Anthropic positions them as complements: Sonnet 5.5 for faster, lower-cost everyday and well-scoped work, and Opus 5.5 for complex tasks that require careful, sustained judgment. Both are part of the same Claude 5.5 family launched in September 2026.
Sources: Anthropic, Claude on X.
About the author
Kai Williams
Kai Williams has been in marketing for years, with a long background in SEO before AEO had a name. He stepped into Answer Engine Optimization the moment AI started reshaping how people search, and has been tracking the shift ever since. At Prompt Insider, he covers AEO, AI marketing, and the future of search, breaking down what is changing and what brands need to do about it.
