Claude

Definition

Claude is a family of large language models developed by Anthropic, an AI safety-focused company founded by former OpenAI researchers. Claude models are designed for conversational assistance, coding, document analysis, and agentic tasks, with Anthropic emphasizing 'Constitutional AI' training methods intended to make the models helpful, harmless, and honest.

Overview

Anthropic was founded in 2021 by siblings Dario and Daniela Amodei along with other former OpenAI researchers, with a mission centered on AI safety research. Claude is the company's flagship product line, accessible via the Claude.ai chat interface, the Anthropic API, and through major cloud platforms including Amazon Bedrock and Google Cloud Vertex AI. Claude models are trained using a technique Anthropic calls Constitutional AI, in which the model is trained to critique and revise its own outputs according to a set of guiding principles ('the constitution'), combined with reinforcement learning from human and AI feedback. This approach is intended to reduce the need for large volumes of human-labeled harmful examples while steering the model toward being helpful, harmless, and honest. Claude has iterated through several generations — commonly organized into size tiers such as Haiku (fastest, most efficient), Sonnet (balanced performance and cost), and Opus (highest capability) — with successive version numbers reflecting improvements in reasoning, coding, long-context handling, and agentic tool use. Claude models support large context windows suitable for analyzing long documents and codebases, and later versions introduced extended/hybrid reasoning modes, computer-use capabilities (allowing the model to control a computer interface), and strong performance on software engineering benchmarks, contributing to Claude's popularity as the underlying model for coding assistants and agentic developer tools such as Claude Code. Anthropic positions safety and interpretability research as core to its mission, publishing work on mechanistic interpretability and AI alignment, and Claude's system prompts and usage policies emphasize avoiding harmful, deceptive, or dangerous outputs.

Key Features

• Developed by Anthropic using Constitutional AI and RLHF-based training
• Offered in multiple size/capability tiers (e.g., Haiku, Sonnet, Opus)
• Large context windows for processing long documents and codebases
• Strong performance on coding and software engineering benchmarks
• Supports agentic tool use and computer-use capabilities in later versions
• Accessible via Claude.ai, Anthropic API, Amazon Bedrock, and Google Cloud Vertex AI
• Emphasis on AI safety, interpretability, and reducing harmful/deceptive outputs

Use Cases

Alternatives

History

Claude is a family of large language models built by Anthropic, an AI safety company founded in 2021 by former OpenAI researchers including siblings Dario and Daniela Amodei. The first Claude models — Claude and the faster, cheaper Claude Instant — were released on March 14, 2023, when Anthropic opened its API. Claude is trained with Constitutional AI, a method in which the model critiques and revises its own responses against a written set of principles (a "constitution") rather than relying solely on human feedback. Claude 2 arrived on July 11, 2023 alongside the public claude.ai chat interface and expanded the context window to 100,000 tokens. The Claude 3 family — Opus, Sonnet, and Haiku, spanning three size and cost tiers — was announced on March 4, 2024, with Claude 3 Haiku becoming generally available on March 13, 2024.

Frequently Asked Questions

Related in AI Models