GitHub Copilot
Infra / DevOps · Freemium
Overview
An AI coding assistant from GitHub and Microsoft, built with OpenAI and now other model providers. It started in 2021 as an autocomplete that suggested whole lines and functions as you type. It has grown into a fuller assistant: a chat panel, an agent that edits across files, a coding agent that works through a whole task on its own, code review, and a command line version. It runs inside popular editors including VS Code, Visual Studio, JetBrains IDEs, Xcode, and Neovim, and on the GitHub website. There is a free tier with limits, paid individual plans, and Business and Enterprise plans for organizations.
What people use it for
The most common use is inline code completion: it suggests the next line or block while you type, and you accept or ignore it. Developers use Copilot Chat to ask questions about code, get an explanation of an error, or draft a function from a description, without leaving the editor. Agent mode takes a written instruction and makes changes across several files, running commands and fixing its own mistakes. The coding agent goes further: you assign it a GitHub issue and it works on a branch and opens a pull request for review. Teams use Copilot code review to get an automated first pass on pull requests. The command line tool answers shell questions and explains commands. Enterprises also use it with their own codebase indexed so answers match house patterns.
Key capabilities
Inline completion and next edit suggestions are included on every paid plan and do not draw from the usage allowance. Copilot Chat, agent mode, the autonomous coding agent, code review, and the CLI are metered. Recent plans use a pool of AI credits measured by token use rather than a fixed request count. Users can choose the model, with higher plans unlocking premium models from Anthropic, Google, and OpenAI. It works in VS Code, Visual Studio, JetBrains IDEs, Xcode, Eclipse, and Neovim, plus GitHub.com and GitHub Mobile. Custom instructions and prompt files let a team steer its style, and it can pull context from a repository or a knowledge base. The free tier gives a capped number of completions and chat messages per month with no card required. Business and Enterprise plans add organization policy controls, pooled credits, content exclusion so chosen files are never used, and audit data, and Enterprise adds SAML single sign on and answers grounded in the organization's own repositories. Plans are Free, Pro, Pro+, Business, and Enterprise, with monthly credit allowances that scale by tier and metered overage beyond them.
Limitations
Copilot inherits the weaknesses of the models under it. It hallucinates APIs and configuration that do not exist, suggests outdated or insecure code, and often misreads the wider context of a project, so its output needs review. The inline completion context is fairly small, which limits how well it understands a large codebase from completions alone. Developers report that quality wobbles when GitHub swaps the underlying models, since the tuning around them does not always keep pace. The 2026 move to token based AI credits made costs harder to predict for heavy chat and agent users. Leaning on it too hard is also linked to weaker code habits.
Insight
Copilot's steadiest value is the boring part: inline completion that finishes the line you were already going to write. That alone is worth the base plan for most developers who code daily, and it is included without eating into the usage credits. The newer agent and coding agent features are more ambitious and more uneven. They can carry a small, well defined task to a pull request, but they also produce confident wrong code, so a human still has to read every line. The shift to token based credits means heavy chat and agent use now has a variable cost that is easy to underestimate. It is deeply tied into GitHub, which is a plus if the team already lives there. Treat the completions as the core product and the agents as a helper that always needs checking.
Pricing
Freemium
Website
Last checked 2026-08-30