What GitHub Copilot is
Copilot launched in 2021 as an autocomplete for code: start typing and it offers the next lines. It has since grown into a set of tools that work in your editor, in the terminal and on GitHub.com. You do not switch editors to use it; it is an extension for the editor you already have.
Copilot is not tied to one AI company. Depending on your plan, you can choose models from OpenAI, Anthropic, Google and others, or let Copilot pick automatically.
Key features
Code completions and next edit suggestions
Inline suggestions appear as you type, from a single line to a whole function. Next edit suggestions predict the next change you are likely to make elsewhere in the file. Paid plans include unlimited completions; the free plan includes 2,000 a month.
Chat and agent mode in the editor
Copilot Chat answers questions about your code, explains errors and writes tests. Agent mode goes further: you describe a change and Copilot edits several files, runs commands and checks the result, asking before risky steps. Custom instructions and MCP servers let you give it your project's conventions and connect it to other tools.
Cloud agent
On paid plans you can assign a GitHub issue to Copilot. It works in a cloud environment and opens a pull request for you to review. On Pro+ and above, you can also hand tasks to third-party coding agents such as Claude and OpenAI Codex (in preview as of October 2026).
Code review
Copilot can review pull requests and file diffs, leaving comments with suggested fixes. It is included from Pro.
Copilot CLI and other places
The Copilot CLI brings chat and agent features to the terminal. Copilot also works on GitHub.com, in GitHub Mobile and in Windows Terminal.
Pricing (as of October 2026)
From GitHub's plans page and plan documentation, checked October 2026. Usage beyond completions is measured in GitHub AI Credits; one credit equals $0.01.
| Plan | Price | Monthly AI credits | Notes |
|---|---|---|---|
| Free | $0 | Limited | 2,000 completions and 50 chat requests a month; auto model selection; agent mode, CLI, MCP. |
| Student | Free (verified) | Allowance | For verified students. |
| Pro | $10/month | 1,500 (1,000 base + 500 flex) | Unlimited completions, cloud agent, code review, choice of models. |
| Pro+ | $39/month | 7,000 (3,900 + 3,100 flex) | Premium models, third-party coding agents (preview). |
| Max | $100/month | 20,000 (10,000 + 10,000 flex) | Priority access to premium models, for heavy agent use. |
| Business | $19/seat/month | 1,900 per user | Organization licenses, policies, usage limits, IP indemnity. |
| Enterprise | $39/seat/month | 3,900 per user | Codebase indexing, Copilot on GitHub.com; requires GitHub Enterprise Cloud. |
Expensive models use credits faster than cheaper ones, so the same allowance covers fewer requests if you pick the top models. Admins on Business and Enterprise decide whether users can go over their allowance and pay for extra usage.
Strengths
- Price. Pro at $10 a month is cheaper than most rival paid plans, and the free plan is usable for light work.
- Works in your editor. Support for VS Code, Visual Studio, JetBrains, Xcode, Eclipse and Neovim means no switching tools.
- GitHub integration. Assigning issues to an agent and getting a pull request back fits naturally into GitHub-based teams.
- Model choice. Several AI companies' models are available in one subscription.
- Business-ready. Policy controls, IP indemnity and no training on Business and Enterprise code make it easy to approve at many companies.
Weaknesses
- Credits are hard to predict. Since usage moved to credits that depend on the model, users report difficulty estimating how far an allowance goes, especially with agents.
- Agent quality varies by editor. The newest features usually reach VS Code first; other editors can lag.
- Can be confidently wrong. Like all AI coding tools, it suggests code that looks right but has bugs or security problems. Review everything.
- Less AI-centric than Cursor. Developers who want AI built into every part of the editor often find Cursor's design more fluid.
- Plan changes. GitHub has reshaped plans and usage rules several times since 2024, which can affect budgets.
Privacy and security
What GitHub states, as of October 2026:
- Code and prompts from Copilot Business and Enterprise are not used to train models.
- Individual plans have settings controlling whether prompts and suggestions are retained and used to improve GitHub's products.
- A filter can block suggestions that match public code, and code referencing shows where matching code came from and its license.
- Business and Enterprise include IP indemnity, and admins can choose which models and features are allowed.
- Content exclusion lets organizations stop Copilot from reading specific files or repositories.
Platforms
- Editors: VS Code, Visual Studio, JetBrains IDEs, Xcode, Eclipse, Neovim
- GitHub.com and GitHub Mobile
- Terminal: Copilot CLI and Windows Terminal
- Operating systems: Windows, macOS and Linux (wherever the editor runs)
Alternatives
- Cursor: an AI-first editor built on VS Code, with deeper agent features and a different pricing model.
- Claude: Claude Code is a strong terminal-based coding agent included with Claude Pro. See Claude vs ChatGPT for how the assistants compare.
- ChatGPT: Codex, OpenAI's coding agent, comes with ChatGPT plans.
Which plan should you pick?
Free is enough to see whether AI suggestions help how you code, and for occasional side projects. The 2,000 completions a month go quickly if you code daily.
Pro is the sweet spot for most individuals: unlimited completions, chat and agent mode with a choice of models, plus the cloud agent and code review. The included credits are enough for moderate chat and agent use, especially if you stick to the standard models.
Pro+ and Max are for people who run agents for long stretches every day, or who want the top premium models as their default. Compare the cost with Cursor and Claude Max if you are in that group.
Business and Enterprise are for organizations. The main reasons to choose them are central licenses, policy controls, usage limits, content exclusion, IP indemnity and the assurance that code is not used for training. Enterprise adds codebase indexing and deeper GitHub.com features for companies on GitHub Enterprise Cloud.
Tips for better results
- Add a custom instructions file to your repository describing your conventions, test commands and structure.
- Keep the files you are working on open; Copilot uses them as context.
- Use cheaper models for routine questions and save premium models for hard problems to stretch your credits.
- Review agent pull requests as carefully as a colleague's.
Who should use GitHub Copilot
Copilot is a good default for individual developers who want solid AI help without changing editors, and for companies on GitHub that need admin controls and legal cover. Start on Free; Pro is the plan most developers need. Consider Pro+ or Max only if you lean heavily on agents and premium models every day.