Overview
- Anthropic released Claude Sonnet 5.5 on Monday as the second member of its Claude 5.5 family and made it available on Anthropic’s platform plus AWS, Google Cloud, and Microsoft Foundry.
- Anthropic reports Sonnet 5.5 generates outputs more than 30% faster than Sonnet 5 and can cut per‑task costs by up to about 30% because it uses fewer tokens for many tasks.
- Company benchmarks show Sonnet 5.5 matching or edging Opus 5.5 on several coding and knowledge tests, including a 70.6% score on Terminal‑Bench 4.0, though Opus remains stronger on complex open‑ended work.
- Independent testers broadly corroborate performance gains but flag higher token use and different cost outcomes at higher effort settings, meaning actual bills will vary by the effort level and workload.
- Anthropic kept Sonnet’s list price unchanged, added Opus‑style cyber safeguards and classifiers to limit misuse, and said a smaller, high‑throughput Haiku 5.5 will arrive in the coming weeks as it positions products for enterprise customers ahead of an IPO.