Anthropic has released Sonnet 5.5, the newest iteration of its mid-tier artificial intelligence model. The company describes the updated model as an ideal assistant for everyday tasks, including office document creation and coding, designed to operate much faster and at a significantly lower cost than its predecessor.
Speed and performance benchmarks
Sonnet 5.5 follows its predecessor, Sonnet 5, which was announced roughly three months prior with a focus on efficient agentic deployment at lower costs than competitors. According to Anthropic, the primary selling point of Sonnet 5.5 is speed. The lab claims the new model is 30 percent faster than the previous version while maintaining a significantly slower rate of token burn.
Within Anthropic’s model hierarchy, Sonnet sits below the Opus model in terms of raw power, though it often provides greater utility in specific circumstances due to its agility. Benchmarks provided by Anthropic indicate that Sonnet 5.5 outperforms Opus 5.5 in agentic coding, a capability attributed to its capacity to spawn multiple agents without exceeding cost thresholds.
Cyber capabilities and upcoming releases
Anthropic reports that Sonnet 5.5 possesses notable cyber capabilities, describing them as comparable to the cyber capabilities found in Opus 5. Consequently, the company states that Sonnet 5.5 is the first model in the Sonnet line to be subjected to the same cyber safeguards previously applied only to Fable and Opus.
Additionally, Anthropic announced plans to release a new version of Haiku, its smallest model, in the coming weeks, though a definitive release date was not specified.
