What is Anthropic?

Anthropic is the AI safety company behind Claude, founded in 2021 by former OpenAI researchers who wanted to build powerful AI more carefully. Siblings Dario and Daniela Amodei run it, and the safety-first framing is not just marketing — the company pioneered constitutional AI and publishes much of its safety research openly.

Its product is the Claude family of models. As of July 2026 the lineup runs from Haiku 4.5 (fast and cheap) through Claude Sonnet 5 — the mid-tier workhorse launched June 30, 2026 — and Opus 4.8, up to the new flagships Claude Fable 5 and Claude Mythos 5. Fable 5 is the version anyone can use; Mythos 5 is restricted to government and select research partners.

The careful reputation has not slowed the business. As of mid-2026 Anthropic had closed a funding round at a reported $965 billion valuation, confidentially filed IPO paperwork, and grown its revenue run rate to a reported $47 billion — roughly neck-and-neck with OpenAI.

If you have chatted with Claude, used Claude Code in a terminal, or called the Claude API, you have used Anthropic. Its core bet is that the most careful lab can also ship the smartest models. So far the market seems to agree.