On September 28, 2026, Anthropic released a new AI model, Claude Sonnet 5.5. It is the second model in the Claude 5.5 family, following Claude Opus 5.5, which launched on September 22.
The announcement
According to the official announcement, Sonnet 5.5 generates output more than 30% faster than the previous-generation Sonnet 5 and lowers the cost per task by up to 30% on many kinds of work. Anthropic attributes the savings to the model completing the same work with far fewer tokens than before.
Anthropic says the model is especially strong at well-defined everyday tasks, bug fixes, and creating documents, slides and spreadsheets.
Key numbers
These are the main figures Anthropic published.
- 70.6% on Terminal-Bench 4.0, an agentic coding evaluation (Sonnet 5: 10.3%)
- 80.1% on OSWorld 2.1, a computer-use evaluation (including partial credit; Sonnet 5: 57.0%)
- 1844 Elo on GDPval-AA v2.1, an evaluation of real-world work tasks (Sonnet 5: 1449)
- API pricing of $2 per million input tokens and $10 per million output tokens (cache reads $0.20, cache writes $2.50)
All benchmark figures come from Anthropic's own announcement.
Where to use it
Sonnet 5.5 is available in the Claude apps and Claude Code, as well as on the Claude Platform for developers (model ID: claude-sonnet-5-5), Amazon Web Services, Google Cloud and Microsoft Azure. In Claude Code and the Claude apps, the "effort" setting, which controls how deeply the model reasons, defaults to Medium. On the Claude Platform for developers, it defaults to High.
Sonnet 5.5 also ships with cybersecurity safeguards against misuse similar to those of Opus 5.5. For high-risk cybersecurity work, it explicitly indicates that it is switching to Sonnet 5.
CoAI's take
This update is mainly about speed and cost, so people who use Claude for everyday work or build with Claude Code are the most likely to notice the difference. Note that the pricing above applies to the API; for how usage limits in the Claude apps and Claude Code may change, check the information for your plan.
