Claude

Anthropic's Claude is a family of dense transformer LLMs trained via Constitutional AI: large-scale pretraining followed by RLHF and RLAIF against an explicit principle set. Models ship in three tiers (Haiku, Sonnet, Opus), support up to 1M token context, and target coding, agentic workflows, and computer use. API access includes Model Context Protocol and tool-use integrations.

Categories