Claude

Claude is a family of large language models (LLMs) developed by Anthropic, designed to be helpful, honest, and harmless. It operates under a constitutional AI framework—following predefined ethical principles to self-evaluate and refine output, reducing bias and enhancing reliability.

Claude excels in handling extended context—some deployments support up to a 1‑million‑token context window—enabling sophisticated tasks like long transcript summarization, multi-document analysis, and deep research.

The model family includes lightweight (Haiku), mid-tier (Sonnet), and powerful (Opus, Fable, Mythos) variants, tailored for tasks from quick answers to complex multi-step reasoning and AI agent orchestration.

• Constitutional AI governance: Claude evaluates its own outputs against ethical guidelines, improving safety, reducing hallucinations, and ensuring alignment with human values.
• Massive context support: Supports contexts up to 1 million tokens (approx. 500 pages), enabling deep analysis of large documents, transcripts, and long-form content.
• Claude Code & Cowork tools: Claude Code reads projects, generates PRs, reviews code, and runs tests. Claude Cowork—a non-technical agent—autonomously executes multi-step tasks in user folders.
• AI agents and MCP integration: Users can create AI agents that plan, act, and collaborate; Model Context Protocol (MCP) enables integration with Microsoft 365 apps like Outlook, Teams, OneDrive, and SharePoint.
• Web search & code execution tools: Built-in tools allow web search within chats and sandboxed code execution, augmenting reasoning and real-world data access.

For developers: Claude Code offers project-level understanding, test execution, PR creation, and multi-agent coordination—ideal for complex coding workflows.

For professionals and knowledge workers: With long context support and document analysis, Claude streamlines research, report drafting, and summarization tasks.

For enterprise users and analysts: AI agents plus MCP integration enable automation across Outlook, Teams, SharePoint, and OneDrive—enhancing productivity in business workflows.

• User signs up via Web, Desktop, or Mobile (iOS/Android) interfaces; must be age 18+ and in supported region.
• User selects model: Haiku (fast/light), Sonnet (balanced), Opus/Fable (deep reasoning).
• Enter prompt naturally in chat, or invoke tools like web search or code execution; model self-evaluates output per its constitution.
• For coding tasks, Claude Code reads repo structure, generates or reviews code, runs tests, and proposes PRs; Routines can schedule tasks on Anthropic’s backend.
• Agents can be configured to perform planning and task execution autonomously (Cowork for non-tech tasks, agent frameworks via MCP).
• Output delivered in chat, with artifacts pane for long content; context memory persists across sessions per plan tier.

Claude offers tiered subscription plans and usage-based API pricing.

• Plans (Apps): Free‑forever tier includes chat, memory, web search, and basic tools; Pro tier upgrades usage, adds Claude Code/Cowork/Design, unlimited projects, Microsoft 365 and Outlook integration; Max tier provides 5× or 20× greater usage, higher output limits, early access to features, priority access during peak usage.
• Team/Enterprise seats: Seat-based pricing with standard and premium levels offering increased usage and organizational controls.
• API pricing: Token-based model billed per million tokens across model variants (Haiku, Sonnet, Opus, Fable/Mythos) with additional costs for prompt caching tiers; web search is charged per thousand searches plus token cost; code execution tool may have separate execution-hour billing.