Anthropic Launches Claude Sonnet 5.5 With Faster Output and Lower Costs per Task

Anthropic has launched Claude Sonnet 5.5, an update to its mid-tier AI model aimed at users who want quick responses for coding and routine office work. The company says the model generates output more than 30% faster than Sonnet 5, which it released about three months ago.

Faster Work Without a Cut to Token Prices

Anthropic is positioning Sonnet 5.5 for well-defined tasks such as fixing bugs and creating documents, presentations, and spreadsheets. Its API prices remain at $2 per million input tokens and $10 per million output tokens, the same as Sonnet 5. The claimed savings come from using fewer tokens to complete a task: Anthropic says that can reduce the cost per task by up to 30%.

The distinction matters for developers running AI agents repeatedly. Even when the price of each token stays the same, a model that finishes work in fewer steps can lower the cost of a workflow.

Coding Results Put Sonnet Closer to Opus

Anthropic reports substantial gains over Sonnet 5 in its evaluations. On Terminal-Bench 4.0, an agentic coding test, Sonnet 5.5 scored 70.6%, compared with 10.3% for its predecessor and 66.4% for Opus 5.5. Results vary across benchmarks, however: Opus 5.5 scored higher on other coding evaluations, and Anthropic says it remains the stronger choice for complex work requiring sustained judgment.

Stronger Cyber Safeguards Accompany Release

Anthropic says Sonnet 5.5’s cybersecurity capabilities have improved enough to warrant safeguards similar to those used for its most capable models. Routine software development remains supported, while some higher-risk cybersecurity requests may fall back to Sonnet 5.

Sonnet 5.5 is available across Anthropic’s platforms and through Amazon Web Services, Google Cloud and Microsoft Azure. Anthropic also plans to add Haiku 5.5, the smaller model in the family, in the coming weeks.