A clodhopper stomps in, types “fix it,” and lands in the soup. A Claudhopper looks first, then leaps. This is the map of the whole Claude pond: every model, every app, the tools that live nearby, and how to get your feet wet on any machine.
No hype. No jargon wall. Some frog puns. We apologize for nothing.
Same pond, same Claude, very different landings. Everything on this page is about moving from the left column to the right one.
| 🥾 The clodhopper | 🐸 The Claudhopper |
|---|---|
| Types “fix it” and hopes. | Says what, where, and what “done” looks like. |
| Pastes the answer straight into production. | Checks where it landed. |
| Uses the biggest model to rename a file. | Picks the frog that fits the pad. |
| Re-explains the whole project in every new chat. | Keeps a memory file and a project. |
| Believes it because it sounded sure. | Asks “how do you know?” |
| Has eleven terminal tabs and no idea which one is on fire. | Runs the agents from one cockpit. |
There are four doorways into Claude. The web app needs zero setup. Claude Code is a terminal tool that can read, write and run code on your own computer — install differs a little by platform, so pick yours. Most people start on the web and move to Claude Code within a week.
Sign up and start chatting. No install, no terminal. Everyday questions, research, writing, image analysis, and quick prototypes through Artifacts.
Open claude.ai →Install the terminal app on Windows. Once in, Claude can read your files, edit documents, organize folders and run code — anything you can do in a shell.
Windows guide →Same idea, fewer steps on macOS — the built-in Terminal handles everything. Five steps from a fresh Mac to your first prompt.
Mac guide →Ubuntu, Debian, Fedora, Arch, WSL, BSD and friends. Install Node, run one command, you're in. Notes for each major package manager.
Linux / Unix guide →Claude isn't a single model — it's a lineup, each tuned for a different trade-off between raw capability, speed and cost. You pick the one that fits the task (and in most tools you can switch mid-conversation). Exact IDs, context windows and prices change often — always confirm the current numbers at platform.claude.com/docs.
Anthropic's Mythos-class flagship, sitting above Opus — frontier reasoning and long autonomous runs with a very large context window. Now the default model in Claude Code, and the one to reach for on the hardest work. The Fable playbook →
The deep-reasoning workhorse: complex coding, multi-step analysis, long agent sessions. The model most builders reach for by default when the job has real teeth.
The sweet spot — fast and sharp enough for most everyday work, at a fraction of flagship cost. Great for drafting, refactors and high-volume calls.
The lightweight — near-instant responses at the lowest price. Built for classification, extraction, routing and anything you run thousands of times.
The same underlying flagship with safeguards lifted in narrow domains, limited to vetted partners. Not generally available — listed here for completeness, not as an option to pick.
The ways you actually touch Claude. Solid tiles ship today; dashed tiles are forward-looking — early-access, recently announced, or on the roadmap. Deep-dive write-ups live on the WholeTech guide; links open there.
Claude doesn't live alone. A growing ring of apps and open-source projects wrap around it — ways to run it, drive it, and plug it into the rest of your stack. These are the ones worth knowing.
An open-source cockpit for coding agents — run Claude Code (and Codex, Cursor, Grok, OpenCode) from one screen, with multi-agent threads, branch-per-thread git and one-button PRs. Drive it from your desktop or your phone. Full install guide for Mac, Windows and Linux.
Anthropic's own terminal agent — reads and writes your files, runs commands, edits across a whole repo, and works from the CLI, your IDE, or the web. The backbone most of this ecosystem is built around.
Wire Claude into Figma, Vercel v0, Cursor, GitHub Actions, n8n and the MCP servers — Drive, Gmail, Calendar, Slack, YouTube — so each tool does what it's best at. Includes chaining recipes.
Claude as a design partner — Artifacts as a sketchbook, the frontend-design plugin, image-in to component-out, and whole design systems. Recipes for heroes, landing pages and themes.
Claude Code installs the same way everywhere once you have Node.js — one npm command. The platform notes below cover the small differences. For the full hand-held walk-through with screenshots, use the per-platform guides linked in Quick start.
# the one command, once Node.js is installed — works on every platform
npm install -g @anthropic-ai/claude-code
# then start it in any project folder
claudeUse the built-in Terminal — no extra shell needed. Install Node (via the installer or Homebrew: brew install node), run the npm command, then claude.
Install Node for Windows, then run the npm command in PowerShell or Git Bash. Claude Code runs great under WSL too if you prefer a Linux shell.
Install Node from your package manager (apt, dnf, pacman, nix…), run the one npm command, you're in. Works on Ubuntu, Debian, Fedora, Arch, BSD and WSL.
Want Claude Code plus other agents on one screen, even from your phone? Install T3 Code on top — it drives the CLI you just set up.
claude and follow the login prompt — your Claude subscription (Pro / Max / Team) authorizes the CLI, no separate API key needed for everyday use. Full docs at code.claude.com/docs.Two separate worlds: subscriptions (a flat monthly fee for the apps and Claude Code) and the API (pay-per-token for what you build). Prices and limits move — these are the shapes, not quotes. Confirm current numbers at claude.com/pricing and platform.claude.com/docs.
The web app with daily limits. Enough to try Claude, ask questions, and build small Artifacts. No card required.
Flat monthly fee, bigger weekly usage, higher-end models, and Claude Code included. Max stacks more usage for heavy daily drivers.
Seats, shared projects, admin controls and higher limits for groups. Enterprise adds SSO, data controls and support.
Billed by input/output tokens, priced per model. Batch is ~50% cheaper for non-urgent jobs and prompt caching cuts repeat-context cost sharply.
Claudhopper points the way; these guides go the distance.
The complete, plain-language guide — products, prompting, design, integrations, and per-platform setup, section by section.
How to get frontier work out of Fable 5 — the mental model, the before/during/after workflow, and the prompting mechanics.
A complete, runnable field guide to standing up an autonomous agent on Claude — nine layers, each ending in a checkpoint you can run.
Anthropic's own reference for the terminal agent — commands, hooks, MCP, settings and IDE integrations.
The source of truth for models, context windows, pricing, parameters and every API surface. Verify specs here.
How a big-footed frog and a Thinking Lantern taught a whole pond to ask better questions. For kids, grandparents, and tired engineers.
Run Claude Code and other agents from one cockpit, including your phone. Install for Mac, Windows and Linux.