Sonnet 4.6 updates Claude’s everyday model

Anthropic introduced Sonnet 4.6 with updates to programming, computer interaction, planning, and work with large documents. The announcement includes a million-token context beta and makes the model the default on several consumer plans. It also describes improvements in following instructions.

Context

Everyday usefulness depends on more than the hardest task a model can solve. Predictability, latency, and the amount of cleanup affect whether people return to it. Repeating familiar tasks on both versions provides a more relevant comparison than assuming every benchmark gain will appear in every workflow.