Product history
The Claude model family
How Anthropic's assistant evolved, and how it connects to Dario Amodei's research agenda.
Claude is the family of AI models developed by Anthropic, the company Dario Amodei co-founded and leads. It is available as a chat assistant, through an API for developers, and in coding tools. This page gives the short history. For current models, always check Anthropic's Claude page.
A short history
Claude launches
Anthropic releases the first Claude to business partners and the public.
Claude 2
A broader release with a larger context window and public availability.
Claude 3 family
Haiku, Sonnet and Opus tiers. See the announcement.
Claude 3.5 Sonnet
A step up in coding and reasoning, introducing Artifacts to the chat interface.
Claude 3.7 Sonnet and Claude Code
A hybrid reasoning model, released alongside a preview of the Claude Code command-line tool.
Claude 4
Claude Opus 4 and Sonnet 4, focused on coding and long-running agentic work. See the announcement.
The Claude 5 family
Newer generations continue to ship. Check Anthropic for the current lineup and model names.
Why Claude matters to the Dario Amodei story
Claude is where Anthropic's research ideas meet users. Techniques like Constitutional AI and commitments in the Responsible Scaling Policy are intended to shape how Claude is trained, evaluated and released. Commercial success also funds Anthropic's safety and interpretability research.
Model tiers
Since the Claude 3 generation, Anthropic has used three names for different trade-offs:
| Tier | Typical role |
|---|---|
| Haiku | Fastest and cheapest; high-volume tasks |
| Sonnet | Balanced capability and speed; everyday work and coding |
| Opus | Most capable; hardest reasoning and agentic tasks |