Anthropic Launches Claude Sonnet 5.5 With 30% Lower Costs Per Task

Anthropic has launched Claude Sonnet 5.5, the second model in its Claude 5.5 family, with improvements in coding, knowledge work and computer use.

Claude Sonnet 5.5 is available across Anthropic's platforms as well as Amazon Web Services, Google Cloud and Microsoft Azure. Developers can access the model through the Claude Platform using claude-sonnet-5-5.

The AI company said Sonnet 5.5 generates outputs more than 30% faster than Sonnet 5 and can cost up to 30% less per task. 

Sonnet 5.5 costs $2 per million input tokens and $10 per million output tokens, while cache reads cost $0.20 per million tokens. Opus 5.5, by comparison, costs $4 per million input tokens and $20 per million output tokens.

Anthropic said Sonnet 5.5 requires fewer tokens per task than Sonnet 5, reducing the overall cost of running workloads despite retaining the same token pricing as its predecessor.

Sonnet 5.5 is positioned as a lower-cost alternative to Claude Opus 5.5 for everyday tasks, bug fixing and creating documents, slides and spreadsheets. Anthropic said Claude Haiku 5.5, which will target high-volume and cost-sensitive applications, will be launched in the coming weeks.

On Terminal-Bench 4.0, which evaluates models on agentic coding tasks, Sonnet 5.5 scored 70.6%, compared with 10.3% for Sonnet 5 and 66.4% for Opus 5.5. 

Anthropic said the improvement also shows up in the cost of running coding tasks. At Medium effort on Terminal-Bench 4.0, Sonnet 5.5 exceeded Sonnet 5’s best score while costing less than one-tenth as much per task.

The model also closed the gap with Opus 5.5 on knowledge-work benchmarks. On GDPval-AA v2.1, Sonnet 5.5 scored 1,844, compared with 1,846 for Opus 5.5 and 1,449 for Sonnet 5.

On OSWorld 2.1, a computer-use benchmark, Sonnet 5.5 scored 80.1%, compared with 81.8% for Opus 5.5 and 57% for Sonnet 5. On Humanity’s Last Exam with tools, it scored 64.5%, against 67.7% for Opus 5.5.

However, Anthropic said benchmark scores capture only part of a model’s capabilities and that Opus 5.5 remains stronger for complex, open-ended tasks that require sustained judgement.

The company has also added cybersecurity safeguards similar to those used for Opus 5.5. Higher-risk cybersecurity requests can fall back to Sonnet 5, while users can continue to use the model for routine software development tasks.