Most capable
Claude Opus
Hardest reasoning, deep analysis, complex agents.
Learn Claude
What is Claude?
Claude is Anthropic's family of large language models, the AI assistants you talk to in plain language to write, analyse, code, and reason. The lineup spans tiers (Opus, Sonnet, Haiku) for different speed-and-capability needs, and extends to Claude Code for agentic coding and the Model Context Protocol (MCP) for connecting Claude to tools and data.
Opus · Sonnet · HaikuClaude CodeMCP
The model family
Three tiers, one family.
Claude comes in tiers so you can trade capability for speed and cost. Each generation refreshes all three.
Balanced
Claude Sonnet
The everyday workhorse, strong and fast for most tasks.
Fastest
Claude Haiku
Low-latency, high-volume work where speed and cost matter.
Who makes Claude
Anthropic, an AI safety company.
Claude is built by Anthropic, founded in 2021 by Dario and Daniela Amodei and other former OpenAI researchers. The model is named after Claude Shannon, and the company's focus on safety and nuance shapes how Claude behaves.
- Where you use it. The claude.ai chat app, an API for developers, and Claude Code in your terminal.
- What it's known for. Long-context reasoning (around a 200K-token window), careful instruction-following, strong coding, and a safety-first design.
- The current generation. The Claude 4 family, refreshed across the Opus, Sonnet, and Haiku tiers.
What it does
What you can do with Claude.
The same model handles very different jobs, which is why "learn Claude" means different things to a writer, an analyst, and an engineer.
- Write and edit long-form content with a feel for tone and nuance.
- Analyse long documents and large codebases in a single context.
- Code end to end, especially through Claude Code's agentic loop.
- Build agents and tools that connect to your systems through MCP.
Pricing
Is Claude free?
Claude has a free tier you can start with today, paid plans for heavier use, and API pricing for developers. Learning to build with Claude on this site costs nothing.
- Free: claude.ai with usage limits.
- Claude Pro: around $20/month for higher limits and the strongest models.
- API: pay per token, priced by model tier (Haiku cheapest, Opus most capable).
Beyond chat
Claude Code and MCP.
- Claude Code is Claude that works in your codebase: reading files, running commands, and editing in a loop, not just chatting. What Claude Code is →
- MCP (Model Context Protocol) is the open standard for connecting Claude to your tools, files, and APIs in a consistent way. MCP, explained →
How to learn Claude
From “what is it” to building with it.
If you just want to use Claude, open claude.ai and start prompting. If you want to buildwith it, and prove it, there's a structured path.
- Start free. The 10-minute diagnostic shows where you stand. No sign-up.
- Work the building blocks. The concepts and production scenarios are the body of knowledge; the six-month curriculum sequences them.
- Prove it. There is an official Claude certification, the Claude Certified Architect (CCA-F). The mock here scores you against the 720 pass bar.
Learning Claude? Get the weekly digest.
One short email a week: what's new in Claude, plus the fastest path to building with it. No account needed.
FAQ
10 questions about Claude
Every Q is phrased as a real Google search query. Answers cite the same evidence-tagged sources used elsewhere on the site.
What is Claude AI?
Claude is Anthropic's family of large language models, the AI assistants you talk to in plain language to write, analyse, code, and reason. The lineup spans tiers (Opus, Sonnet, Haiku) for different speed-vs-capability needs, and extends to Claude Code (agentic coding) and the Model Context Protocol (MCP) for connecting Claude to tools and data.
Is Claude free?
Claude has a free tier at claude.ai with usage limits, plus paid plans and API access for heavier use. Learning to build with Claude is free here: the diagnostic, concepts, and a mock for the architecture certification cost nothing.
Claude vs ChatGPT: what's the difference?
Claude is Anthropic's model; ChatGPT is OpenAI's; Gemini is Google's. All three are frontier assistants. Claude is widely used for long-context work, coding (via Claude Code), and a safety-first design approach. The right one depends on your task. This site is about building well *with Claude* specifically.
What is Claude 101?
"Claude 101" is shorthand for getting started with Claude: what the models do, how to prompt them, and where Claude Code and MCP fit. This page is a plain-English version; the [concepts](/concepts) pillar goes a level deeper into the building blocks.
Are there Claude tutorials or courses?
Yes. Anthropic publishes learning material, and there are community tutorials and courses. If your goal is to build production systems with Claude, the fastest structured path is the [six-month curriculum](/curriculum) here, which sequences the concepts and scenarios behind the certification.
What is Claude Code?
Claude Code is Anthropic's agentic coding tool: Claude that can read your repo, run commands, and edit files in a loop, not just chat. See [what Claude Code is](/knowledge/claude-code-101) and how its skills map to the certification.
Is there a Claude certification?
Yes. The official one is the Anthropic Claude Certified Architect (CCA-F). See [the Claude certification](/claude-certification) and the [exam guide](/exam-guide), and try the [free diagnostic](/practice/diagnostic).
Who makes Claude?
Anthropic, an AI safety company founded in 2021 by Dario and Daniela Amodei and other former OpenAI researchers. Claude is named after Claude Shannon, the founder of information theory, reflecting Anthropic's research-and-safety orientation.
How much does Claude cost?
Claude has a free tier at claude.ai with usage limits, a Claude Pro plan (around $20/month) for higher limits and the strongest models, and API pricing per token for developers. Claude Code is available through paid plans and the API. Learning to build with Claude on this site is free.
How large is Claude's context window?
Claude offers one of the largest context windows available (on the order of 200K tokens), which is why it is widely used for whole-codebase analysis and long-document reasoning. Managing that context well is part of what the Claude certification tests (domain D5).