Anthropic launches faster and cheaper Sonnet 5.5 model

Anthropic has introduced Sonnet 5.5, the latest iteration of its mid-tier artificial intelligence model. Arriving roughly three months after the debut of Sonnet 5, this updated version focuses heavily on processing speed and cost efficiency. The company claims that Sonnet 5.5 operates 30 percent faster than its immediate predecessor and features a significantly slower token burn rate, making it a highly economical option for high-volume tasks.

Though positioned below the flagship Opus model in Anthropic's hierarchy, Sonnet 5.5 outperforms the anticipated Opus 5.5 on agentic coding benchmarks. This advantage stems from the model's agility, which allows developers to spin up multiple agents simultaneously without triggering prohibitive cost thresholds. Additionally, the new model boasts cybersecurity capabilities comparable to Opus 5. Because of these advanced security thresholds, Sonnet 5.5 is the first model in the Sonnet line to be integrated with the strict safety protocols previously reserved for the Fable and Opus models.

For software engineers and enterprise practitioners, these upgrades translate to a more viable tool for everyday workflows, such as writing code and generating complex office documents. The combination of rapid execution and reduced token consumption allows teams to run complex, multi-agent systems that were previously too expensive or slow to deploy at scale.

The release intensifies the ongoing competition among major AI laboratories. Anthropic plans to follow this release with an updated version of Haiku, its smallest model, in the coming weeks. This rollout follows recent moves by rivals, including OpenAI's launch of its updated Sol and Luna models, and Meta's introduction of a new model designed to power its smart glasses.