AI pair programmer from GitHub. Inline completions, multi-line suggestions, slash commands, chat for explaining and refactoring, and agent mode that can author PRs end-to-end. Trained on public code with enterprise filters for license-safe output. Available in VS Code, JetBrains, Neovim, Visual Studio, and the GitHub web UI.
No pricing data on file.
The default AI coding assistant for teams already standardized on GitHub. The 2025–2026 generation is no longer inline completion alone; Copilot Chat, Workspace, and the autonomous Copilot Coding Agent now cover the lifecycle from PR description generation to multi-file edits triggered directly from an issue. Per-task model choice across GPT-5, Claude Sonnet, and Gemini is the recent inflection — engineers can swap models without leaving the IDE, and the agent runs jobs in GitHub-hosted sandboxes that produce real PRs.
High fit for organizations on GitHub Enterprise where centralized policy, audit logs, and IP indemnification matter more than raw model quality. Individual contributors tolerate it; engineering managers like the per-seat predictability and the SSO story for procurement.
Watch out for: feature parity across IDEs is uneven — VS Code gets new capabilities first, JetBrains and Neovim trail by months, and Vim users live in a different product entirely. The autonomous agent occasionally proposes changes that bypass repo conventions because it does not always read CONTRIBUTING.md or AGENTS.md. Per-seat pricing scales linearly even when half the licensed developers barely use it.
