Overview
- Anthropic announced Sonnet 5.5 as the latest mid-tier Sonnet model and says it focuses on everyday assistant tasks such as coding and document work.
- The company reports Sonnet 5.5 runs about 30% faster than Sonnet 5 and uses noticeably fewer tokens, a change it says will lower per-request operating costs.
- Internal benchmarks published by Anthropic show Sonnet 5.5 outperforming Opus 5.5 on multi-agent coding tasks, a result the company attributes to cheaper multi-agent deployments that avoid cost caps.
- Anthropic says Sonnet 5.5 is the first Sonnet to adopt the same network offense/defense protections used by its Opus and Fable models, bringing higher-tier security controls to a lower-cost option.
- The company said a smaller model called Haiku will arrive in the coming weeks and framed Sonnet 5.5 as part of a rapid industry update cycle; all performance and benchmark claims reported here come from Anthropic and have not been independently verified.