Anthropic Claude Sonnet 5.5

Anthropic released Claude Sonnet 5.5 as the direct successor to Claude Sonnet 5, continuing the Sonnet-class line aimed at well-scoped everyday tasks. Third-party hosts describe it as Anthropic's most capable Sonnet model to date, with thinking always on and effort serving as the main lever for trading depth, latency, and cost. That positioning makes the model a fit for coding workflows such as building features and fixing bugs, as well as for producing polished documents, slides, and spreadsheets where clear communication matters.

Snowflake's launch coverage points to a benchmark jump, reporting that Anthropic scored 70.6% on Terminal-Bench 4.0 with Sonnet 5.5, an almost sevenfold gain over the prior generation, underscoring its strength in real software engineering tasks. The model is being rolled out across multiple platforms around its release, including public preview through Snowflake Cortex Inference and listing on the OpenRouter unified API alongside a Sonnet 5.5 batch variant and Claude Opus 5.5, which signals broad third-party ecosystem support for teams that want to keep work inside their existing data and developer environments.