What is Claude?
Claude is a family of AI assistants built by Anthropic, designed around careful reasoning, long-form writing, and software engineering. You talk to it at claude.ai, in desktop and mobile apps, through the Anthropic API, or from the terminal with Claude Code.
The lineup splits by speed and depth: Haiku is the fast, cheap model for high-volume work; Sonnet is the balanced default; Opus is the deepest reasoner for hard problems. All three handle text, images, and documents, and connect to external tools and data through MCP (the Model Context Protocol Anthropic created).
If you mostly care about thinking quality, writing that doesn't read like a template, and coding that actually runs, Claude is usually the model to reach for first.
What it's best for
- Software engineering: reading large codebases, writing and debugging code, and running multi-step agentic tasks (Claude Code is a terminal agent that edits your repo directly).
- Long-form and nuanced writing: drafts, rewrites, and edits that keep your voice instead of flattening it.
- Working over long documents (contracts, RFCs, research papers, transcripts) within a large context window.
- Structured analysis where being careful and admitting uncertainty matters more than sounding confident.
- Tool use and automation via the API and MCP: connecting the model to your own data, systems, and actions.
- Artifacts: generating runnable code, documents, or small apps in a side panel you can iterate on.
Where it falls short
- Native image, audio, or video generation. Claude reads images but does not create them. Use a dedicated image/video model for that.
- Real-time information by default. Without a connected search tool it answers from training data, not today's news.
- The cheapest possible API bill at massive scale. Open-weight models you self-host can undercut it.
Getting started
Sign up at claude.ai with an email or Google account and you can start chatting on the free tier immediately. Install the desktop app (macOS/Windows) or mobile app for the same account everywhere.
Pick the model in the composer: Sonnet is the sensible default. Switch to Opus for genuinely hard reasoning, or when a first answer wasn't deep enough.
Projects, Artifacts, and files
Projects let you give Claude persistent context: drop in reference docs, style guides, or a codebase summary, and every chat in that Project starts already knowing them.
Artifacts render code, documents, or small interactive apps beside the conversation, so you can refine an output in place rather than copy-pasting from the chat. Upload PDFs, images, spreadsheets, and code files directly into a message for analysis.
Claude Code and the API
Claude Code is a command-line agent that reads, edits, and runs code in your own repository, useful for refactors, test-writing, and multi-file changes. It's the same engine behind much of Claude's coding reputation.
For building software on top of Claude, the Anthropic API exposes the models plus prompt caching, tool use, and MCP connectors. This is the path to embedding Claude in your own product or internal tooling.
Getting better answers
Give Claude a role, the constraints, and an example of the output you want, and it follows structure well. For hard problems, ask it to think step by step or to critique its own first draft.
When accuracy matters, ask it to flag what it's unsure about. Claude is comparatively willing to say 'I don't know,' which is a feature, not a bug, for production work.
Privacy and your data
On the consumer plans, Anthropic does not use your conversations to train its models by default. The API, Team, and Enterprise tiers are built for work data, with stricter retention and admin controls; Enterprise adds SSO and audit logging.
For regulated or confidential material, prefer the API or a business plan over a personal account, turn off any history features you don't need, and keep secrets out of prompts. The same hygiene applies to any third-party service.
What Claude costs
Approximate, in USD, as of January 2026. Prices change often. Confirm on the official site before you rely on them.
Free
$0
Daily message limit on the current Sonnet model; web, desktop, and mobile.
Pro
~$20 / month
Higher limits, access to Opus, Projects, and more usage of Claude Code.
Max
~$100 to 200 / month
Much higher usage ceilings for heavy daily users; two tiers.
Team / Enterprise
~$25 to 30 / user / month +
Central billing, collaboration, and admin controls; Enterprise adds SSO and more.
API
Usage-based
Priced per million input/output tokens, by model; prompt caching cuts repeat costs.
Example prompts
Copy these into Claude as starting points, then adapt them to your task.
Refactor with a safety net
You are a senior engineer. Here is a file. Refactor it for readability without changing behavior. List every change you made and why, and flag anything you're unsure preserves the original behavior.
Edit, don't rewrite
Tighten this text for clarity and flow. Keep my voice and structure, and don't make it sound generic. Show the edited version, then a short bullet list of what you changed.
Reason carefully
Think step by step before answering. Lay out your assumptions, work through the problem, then give a final recommendation. Mark anything you're uncertain about.
Review a document
Read the attached document and give me a one-paragraph summary, the three most important points, and any claims that aren't supported by the text itself.
Claude
common questions.
Direct answers to the questions we get asked the most. If yours isn't covered, write to the team.