Constitutional AI framework. Instead of relying on hidden human feedback, Claude evaluates its own responses against a "constitution" — a list of rules like Don't generate harmf

Claude 4 Sonnet is a leaner model, with improvements built on Anthropic's Claude 3.7 Sonnet model. The 3.7 model often had problems with overeagerness and sometimes did more than the person asked it to do, Amodei said. While it's a less resource-intensive model, it still performs well, he said.

Anthropic says Claude Opus 4 is its most powerful model and the best coding model in the world, while Sonnet 4 is replacing Sonnet 3.7 in the chatbot.