Anthropic releases Claude Sonnet 5.5, a faster Sonnet that costs less for most work
Anthropic has released Sonnet 5.5, the second model in the Claude 5.5 family. According to the company, it runs faster than Sonnet 5, uses fewer tokens for the same work, and its price has not changed.
Anthropic released Claude Sonnet 5.5, the second model in the Claude 5.5 family, on September 28, 2026. According to the company's announcement, the new model is a clear upgrade over Sonnet 5: it runs faster and costs less for most work. Its price is the same as Sonnet 5, but the company says it usually needs far fewer tokens to do the same job. Anthropic calls Sonnet 5.5 its fastest Sonnet model to date.
The company positions the model as a faster, lower-cost complement to Opus 5.5, which came out last week. According to Anthropic, Opus 5.5 is built for complex work that needs careful judgment, while Sonnet 5.5 is meant for well-scoped everyday tasks such as fixing bugs and creating polished documents, slides and spreadsheets. The announcement also says Sonnet 5.5 writes more clearly than the previous generation and has a sharp eye for design. Haiku 5.5, the third model in the family, is expected in the coming weeks.
Close to Opus 5.5 in tests
According to the table Anthropic shared, Sonnet 5.5 comes out ahead of Opus 5.5 on Terminal-Bench 4.0, an agentic coding test, and trails it slightly on most of the other tests. On knowledge-work tests it is nearly level with Opus 5.5. The company says that even at low and medium effort, Sonnet 5.5 beats Sonnet 5's best score on several tests at a much lower cost. Anthropic adds that benchmark scores capture only one side of a model and that Opus 5.5 remains clearly stronger at open-ended work that needs sustained judgment. The footnotes say the model scored lower at its highest effort setting on one coding test, and that two tests by the independent group Artificial Analysis ran on a pre-release version with a bug that has since been fixed.
Where it is available
Sonnet 5.5 is available in the Claude apps, in Claude Code and in the API as claude-sonnet-5-5, and also through Amazon Web Services, Google Cloud and Microsoft Azure. According to the documentation, it accepts inputs of up to 1 million tokens. The default effort is medium in Claude Code and the apps, and high in the API. The documentation lists the changes developers need to watch when moving from Sonnet 5. Sonnet 5 remains available. Current prices are on Anthropic's site.
According to Anthropic, Sonnet 5.5 is the first Sonnet model to launch with cyber safeguards; for higher-risk cybersecurity requests it visibly falls back to Sonnet 5. In the company's behavioral audits, the model matches or improves on Sonnet 5 on most measures of alignment, misuse resistance and honesty. Anthropic still notes that the model may have tendencies it has not found, and that some microbiology and virology requests may be flagged in error. For a fuller picture, it is worth following evaluations from independent researchers as well.
Sources: Anthropic: Introducing Claude Sonnet 5.5, Claude docs: Sonnet 5.5, Anthropic: Sonnet 5.5 system card.
