Anthropic's Sonnet 5.5: Near Opus Speed, Lower Cost

Anthropic has released Claude Sonnet 5.5, an enhanced version of its mid-tier AI model. This update significantly boosts speed and efficiency, with output generated over 30% faster and total task costs reduced by up to 30% compared to its predecessor, Sonnet 5.

Sonnet 5.5 demonstrates performance remarkably close to Anthropic's flagship model, Opus 5.5, on various evaluations. It shows particular strength in coding tasks, achieving competitive scores on benchmarks like Terminal-Bench 4.0 and CursorBench 4.0. This closer parity means enterprises have a more powerful, cost-efficient option that may reduce the need for top-tier models.

The new model is optimized for common business tasks, including software debugging, coding, and document generation. While Opus 5.5 remains the preferred choice for complex, open-ended work, Sonnet 5.5's efficiency gains are substantial. Early customer tests from companies like Box and Zendesk confirm faster processing times and improved accuracy.

Furthermore, Sonnet 5.5 integrates advanced cybersecurity safeguards previously reserved for higher-tier models. These enhancements aim to protect against various cyber threats and model distillation attacks. Anthropic is also expanding its Cyber Verification Program to allow approved defenders access to more robust capabilities across its model range.