Anthropic

Anthropic builds the Claude models and positions itself as a safety-focused frontier lab. It was founded in 2021 by siblings Dario Amodei (CEO) and Daniela Amodei (President), with several other former OpenAI researchers.

What Anthropic is known for

• Claude: a family of assistants and API models used heavily for coding, agents and enterprise work.
• Claude Code: an agentic coding tool that became one of the most widely used developer SI products.
• Safety research: Constitutional AI, interpretability and alignment research, and the first Responsible Scaling Policy (2023).
• Model Context Protocol (MCP): an open standard for connecting models to tools and data, released in 2024. See tool use.

Current models (October 2026)

Model	Launched	Price (in / out per 1M tokens)	Positioning
Claude Fable 5.1	September 1, 2026	$10 / $50	Demanding reasoning, long-horizon agents
Claude Opus 5.5	September 22, 2026	$4 / $20	Default for most workloads
Claude Sonnet 5.5	September 28, 2026	$2 / $10	Speed and intelligence balance
Claude Haiku 5.5	October 7, 2026	From $0.10 / $0.50 (higher above 100K-token prompts)	High-volume, latency-sensitive tasks

All four 5.x models offer 1-million-token context windows and up to 128K output tokens.

Policy

Dario Amodei signed the White House Accord on Super Intelligence on September 29, 2026. Anthropic has been more open than most labs about its timelines. In 2025 it told the White House it expected “powerful AI” by late 2026 or early 2027. See When will superintelligence arrive? It has also disclosed safety incidents involving its models’ agentic behaviour, part of the backdrop to the Accord.