Claude Sonnet 5.5 is a meaningful upgrade for teams that use AI in everyday development and document workflows. Anthropic announced it on September 28, 2026, as the second model in its Claude 5.5 family. The company reports faster output and better results on selected coding and knowledge-work evaluations than Sonnet 5, without raising standard token rates.
The practical question is where Sonnet 5.5 passes your quality threshold with less waiting or fewer tokens—not whether every task should move away from Opus 5.5.
Claude Sonnet 5.5 at a glance
| Item | Anthropic’s published detail |
|---|---|
| Claude Platform model ID | claude-sonnet-5-5 |
| Input | $2 per million tokens |
| Output | $10 per million tokens |
| Cache reads | $0.20 per million tokens |
| Cache writes | $2.50 per million tokens |
| Availability | Anthropic platforms, including AWS, Google Cloud, and Microsoft Azure |
Anthropic says Sonnet 5.5 generates output more than 30% faster than Sonnet 5. It also reports up to 30% lower cost per task in its tests because the model often uses fewer tokens. That is not a 30% reduction in the listed token prices, which remain the same as Sonnet 5.
What changed?
More capable coding and focused work
Anthropic reports large gains on its selected terminal and coding-agent evaluations, along with improvements in computer use and professional knowledge work. Early testers describe faster iteration on bug fixes, codebase understanding, documents, slides, and spreadsheets. These are provider and tester reports, not independent guarantees for a different codebase or workflow.
Anthropic still positions Opus 5.5 as stronger for complex, open-ended work requiring sustained judgment. A sensible evaluation is to route well-scoped tasks to Sonnet, retain Opus or a human escalation for uncertain cases, and compare cost per accepted outcome rather than cost per token alone.
A migration detail that can break an integration
Developers can use claude-sonnet-5-5 on the Claude Platform. Anthropic says that if your Sonnet integration currently runs with thinking off, you need to move to the new between_tools setting before adopting Sonnet 5.5. The official migration guide also identifies settings that now return errors, including forced tool choice. Test tool calls and existing conversation state; do not assume a model-ID swap is sufficient.
Anthropic also says preserved thinking is expanding. This can matter when conversations are moved between accounts, including account switching mid-session. Check the provider’s preserved-thinking documentation if your application transfers conversation history.
Safety and limitations
Anthropic says routine software development remains available, while some higher-risk cybersecurity requests visibly fall back to Sonnet 5. It also describes biology safeguards and anti-distillation controls. Its behavioral audits show improvements on most measured dimensions, but the company acknowledges that no evaluation catches every failure.
Keep tool permissions narrow, verify outputs, and require explicit approval for production changes, external messages, access changes, or difficult-to-reverse actions. A faster model does not remove your review and release process.
How to test Sonnet 5.5 in a real product
Choose several representative tasks: a contained bug fix, a multi-file change, a document or data-extraction task, and one failure-prone workflow. Compare Sonnet 5.5 with your existing model on acceptance quality, latency, total tokens, tool calls, retries, human correction time, and total cost per completed task. Include edge cases and failed attempts in the results.
For developers using a cloud provider, confirm the exact provider, region, quota, and model identifier before rollout. Availability across Anthropic, AWS, Google Cloud, and Azure does not mean every account has identical deployment settings.
How I can help you use it
Through AI consulting and custom AI development, I help teams choose models, implement APIs, design evaluations, and add safe approval boundaries to agent workflows. I can also integrate AI into web applications, mobile apps, and workflow automation.
Official sources
Frequently asked questions
What is Claude Sonnet 5.5?
Claude Sonnet 5.5 is Anthropic's September 2026 update to Sonnet 5, aimed at coding, well-scoped everyday work, and document creation. Anthropic positions Opus 5.5 for more complex, open-ended judgment.
How much does Claude Sonnet 5.5 cost?
Anthropic lists $2 per million input tokens, $10 per million output tokens, $0.20 per million cache-read tokens, and $2.50 per million cache-write tokens. These are the same standard token rates as Sonnet 5; total task cost varies.
What is the Claude Sonnet 5.5 API ID?
Anthropic lists the Claude Platform model ID as claude-sonnet-5-5. Check provider and region availability before changing a production integration.
Is there a migration change for thinking-off integrations?
Yes. Anthropic says integrations running Sonnet with thinking off must switch to the new between_tools setting before moving to Sonnet 5.5. Review its official migration guide and test existing conversations and tool flows.