Claude is Anthropic's family of safety-focused language models for coding, agents, reasoning, and knowledge work. Claude Opus 5 is the current Opus generation.
Claude is Anthropic's flagship model family and was originally trained using the company's Constitutional AI feedback method. Anthropic introduced the Claude 3 Haiku, Sonnet, and Opus tiers in 2024, added extended thinking with Claude 3.7 in 2025, and launched the Claude 4 generation for coding and agent workflows in May 2025. Claude Opus 5, released in July 2026, adds a one-million-token context window, thinking by default, deeper test-time reasoning, and stronger long-horizon coding, multi-agent, vision, and enterprise-work capabilities.