Anthropic

Overview

Anthropic is an American AI safety company founded in 2021 by former OpenAI VP of Research Dario Amodei, his sister Daniela Amodei, and several other ex-OpenAI researchers. The company was incorporated as a public-benefit corporation, with safety as a primary fiduciary duty.

The Claude family

Anthropic's flagship product is Claude, a family of large language models known for long-context handling, careful reasoning, and adherence to safety guidelines via "Constitutional AI" — a technique in which models are trained against a written set of principles rather than purely human feedback.

• Claude 1 (Mar 2023) — initial release
• Claude 2 (Jul 2023) — 100K-token context window
• Claude 3 family (Mar 2024) — Haiku, Sonnet, Opus tiers
• Claude 3.5 Sonnet (Jun 2024) — broke ahead of GPT-4o on coding
• Claude Sonnet 4.6 (2025) — current state-of-the-art on SWE-bench Verified (72.7%)
• Claude Opus 4.7 (2026) — frontier reasoning model

Investment & partnerships

• Google invested ~US$2 billion in 2023
• Amazon committed up to US$8 billion across 2023–2024
• Workloads run primarily on AWS Trainium and Google TPU infrastructure
• Backed Salesforce, Menlo Ventures, Spark Capital

Safety research

Anthropic publishes substantial AI alignment research, including work on mechanistic interpretability, sleeper agents, and Constitutional AI. The company maintains a Responsible Scaling Policy that gates model deployment on safety evaluations.