Better default outcomes with Claude Sonnet 4.5 as the standard model

Claude-based agents now default to Claude Sonnet 4.5, improving baseline reasoning quality and response consistency without any configuration changes. This upgrade delivers better out-of-the-box performance while remaining fully configurable if you need to pin an earlier model for reproducibility.

Details

Improved default quality and tool-use reliability
No code changes required; override the model if you need prior behavior

Who this is for: Teams using Anthropic models who want better results by default.