Anthropic launches Claude Sonnet 5.5 with 30% faster output and lower per-task costs

Anthropic has released Claude Sonnet 5.5, the second model in its Claude 5.5 family after last week’s Opus 5.5. The company says it generates output more than 30% faster than Sonnet 5 and costs up to 30% less per task. The savings come from using fewer tokens and tool calls, rather than cheaper individual tokens.

What Claude Sonnet 5.5 is built to handle

Sonnet 5.5 targets everyday work such as coding, debugging, and building polished documents, slides, and spreadsheets, with a sharp eye for design. The performance jump is particularly noticeable in coding. On Terminal-Bench 4.0, a coding test, it scores 70.6% versus Sonnet 5’s 10.3%, and chart reading jumps from 15.6% to 61.6%.

Sonnet 5.5 nearly matches Opus 5.5 on the GDPval-AA knowledge-work benchmark, though Anthropic says Opus 5.5 still leads on complex, open-ended work needing sustained judgment. Early testers backed up the speed claims, with Box reporting runs 2.4x faster and Zendesk seeing tickets processed 20% faster.

RELATED COVERAGE:

Same token prices but fewer tokens per task

The API pricing itself has not changed. Sonnet 5.5 costs $2 per million input tokens and $10 per million output tokens, the same as Sonnet 5 and OpenAI’s GPT-6 Sol. The savings come from using fewer tokens and steps. Lovable reported about a third fewer tool calls in coding tests.

Lower effort settings can beat Sonnet 5’s best scores on some tests for about a tenth of the cost per task. You’ll also see Medium effort by default in Claude apps. These figures come from Anthropic, and independent testing hasn’t confirmed them yet.

Sonnet 5.5 is also the first Sonnet model with cybersecurity safeguards, so higher-risk security requests fall back to Sonnet 5. It’s live on AWS, Google Cloud, and Azure, and Haiku 5.5 arrives in the coming weeks.