Getting Started with GitHub Copilot

An AI pair-programmer that lives inside your code editor, suggesting lines and answering questions as you work.

Current prices: Free, from $10/mo · See all plans

Last reviewed 2026-10-14 · for GitHub Copilot

How it works

Copilot watches the code around your cursor and suggests what probably comes next — a line, a function, a block — which you accept with a keystroke or ignore by carrying on typing. Alongside that, you can ask it things in plain English in a chat panel and get answers grounded in the code you have open.

What it is

GitHub Copilot is an AI coding assistant that runs inside your editor — it autocompletes code as you type, and answers questions in a chat panel about the file or project you're in. It's built by GitHub with models from several AI labs, and it works in the popular editors (VS Code, Visual Studio, the JetBrains IDEs, and Vim/Neovim) rather than being its own separate app. It's aimed at anyone who writes code, from first-project beginners to professionals; you don't need to be an expert to get value from the autocomplete.

Before you start

There's a Free plan (limited chat and agent usage; GitHub no longer publishes a single completion-count we could confirm on 2026-10-14). Paid individual plans step up through Pro, Pro+ and Max, plus a GitHub AI Credits layer; current prices are on its Launchpad page. Verified students and popular open-source maintainers can still get paid features free.

Install it from the official source only — the Copilot extension inside your editor's built-in marketplace, or via github.com. Don't install a "Copilot" plugin from an unofficial third-party listing.

You need a GitHub account and one of the supported editors. If you don't already have an editor, VS Code is free and the most common starting point.

Security & privacy

Copilot sends the code around your cursor, plus your chat messages, to GitHub's servers to generate suggestions. On a work codebase, check your employer's policy before turning it on.

The data-training default worth knowing

On the individual plans (Free, Pro, Pro+), GitHub may use your interactions — including your inputs, the suggestions you get, code snippets, and surrounding context — to train and improve its models unless you opt out in settings.
The Business and Enterprise plans are excluded from that training by default.
If you're on an individual plan and don't want your code used for training, the opt-out is a settings toggle — flip it before you start on anything sensitive.

Suggestions can occasionally mirror public code, and generated code can carry bugs or security flaws. Treat every suggestion as a draft to review, not a finished answer — you're still the author on the hook for what ships. (The paid Business/Enterprise tiers add IP indemnity; the individual tiers don't.)

What it's good (and not good) at

Best at

Autocompleting boilerplate and repetitive code you'd otherwise type by hand
Explaining unfamiliar code in plain language
Drafting tests, comments, and small functions from a description
Keeping you in flow without tab-switching to a separate chatbot

Okay at

Larger multi-file changes via agent mode (workable, but review closely)
Niche or less-common languages (quality tracks how much public code exists)
Debugging — it helps, but can confidently suggest wrong fixes

Avoid

Trusting generated code without reading it
Security-critical or licensing-sensitive code without human review
Treating it as a substitute for understanding what your code does

First things to try

Pick your editor and install the extension

Install Copilot from the official marketplace inside VS Code (or your JetBrains IDE / Visual Studio), then sign in with your GitHub account.

Why this one: Everything else depends on this; the official in-editor marketplace is the safe source.

Tweak: No editor yet? Install VS Code first — it's free.

Start on the free tier

Choose Copilot Free to try autocomplete and limited chat with no card on file before deciding on a paid plan.

Why this one: You can judge whether it fits your workflow at zero cost and zero commitment.

Tweak: Student or OSS maintainer? Check eligibility for free paid-tier access.

Set your training preference

In Copilot settings, decide whether GitHub may use your interactions to train its models, and toggle the opt-out if you'd rather it didn't.

Why this one: On individual plans this is on by default; setting it deliberately before you code is the one thing not to skip.

Try it on real code

Write a function that takes a list of numbers and returns the average. Then write three tests for it.

Why this one: Uses both halves of Copilot at once — the autocomplete drafts the function, chat drafts the tests — so you see how it actually feels.

Tweak: Swap in a small task from something you're actually working on.

Common pitfalls

Credits ≠ autocomplete. Inline completions are unlimited on paid plans; only chat/agent/review draw down credits. Heavy agent use is what burns the monthly allowance.
The free tier is a taste, not a workflow. Its monthly caps are for evaluating, not for daily reliance.
The training opt-out is per-account and easy to miss. If your code is sensitive, set it before your first session, not after.

Where to next

AI Coding & Development Local & Open-Source AI

Want the full picture first? See the GitHub Copilot overview for pricing, the free tier and related tools.