Anthropic has released Claude Sonnet 4.6, a new version of its AI model that aims to approach the performance of its high-end range, Opus, but with a focus on cost and availability for daily use. The update, which maintains the same price, is described with improvements in programming, reasoning with extensive context, and office work. It is established as the default model for Free and Pro users.
Technical improvements in programming and agent capabilities ?™ï?
The update focuses on key technical skills. Anthropic notes advances in programming tasks and computer use, which includes smoother interaction with tools and development environments. Work has also been done on agent planning and reasoning over long contexts, aspects useful for analyzing extensive code or complex documents. The model is already operational in the API, Claude Code, and major cloud platforms.
Opus stays home, Sonnet goes to the party (of code) ??
It seems the strategy is clear: let the premium model, Opus, dedicate itself to philosophical problems in its ivory tower, while Sonnet 4.6 is the one that really gets its hands dirty with our messy code and office documents. It's the functional brother who arrives at the meeting on time and without complaining, ready to debug that script at 11 at night. A true workhorse, although it might dream of being something more than the practical model in the family.