What Is GitHub Copilot? Plans, Models, and the Full Platform
- Describe the three interaction modes in GitHub Copilot and what each is suited for
- Identify which models are available in Copilot and how to switch between them
- Compare the current Copilot plans and explain how usage-based billing works
More Than an Autocomplete Tool
GitHub Copilot is Microsoft and GitHub's AI coding assistant. It lives inside your IDE, on GitHub.com, and in your terminal. When developers first tried Copilot in 2021, it completed lines and suggested the next function. By 2026, it plans multi-step tasks, runs terminal commands, opens pull requests autonomously, and reviews your colleagues' code.
Understanding the full platform before you start is important. There are three distinct interaction modes, and each is suited to a different kind of task. Mixing them up — using chat when you should be using inline completions, or using agent mode when a quick edit would do — slows you down rather than helping.
The Three Interaction Modes
At the IDE level, GitHub Copilot gives you three ways to interact:
- Inline completions — Ghost text appears as you type, right in your editor. Press Tab to accept, Escape to dismiss. This is the fastest path from intent to code when you already know what you want to write.
- Copilot Chat — A conversation panel with three sub-modes: Ask (questions and explanations), Edit (targeted refactors you review as a diff), and Agent (autonomous multi-step execution with tool use and terminal commands).
- Copilot Coding Agent — A fully asynchronous cloud agent. You assign a GitHub Issue, and Copilot works independently inside GitHub Actions — no IDE required — then opens a draft pull request for your review.
These are not the same feature at different power levels. They solve different problems, and you will use all three regularly once you know when to reach for each one.
Multi-Model: You Choose the Brain
Copilot is not tied to a single model. Per-query model switching lets you pick the right model for the job:
- GPT-4o and GPT-4.1 (OpenAI) — Fast, strong general-purpose coding
- Claude Sonnet and Opus (Anthropic) — Strong for long-context work and nuanced reasoning
- Gemini models (Google) — Alternative for certain tasks
The model dropdown appears in the Copilot Chat panel. Inline completions use the default completion model, which GitHub optimises for speed.
Plans and Pricing (2026)
Usage-Based Billing
As of June 2026, GitHub Copilot uses AI credits for agentic features. Code completions and Next Edit Suggestions do not consume credits — they are unlimited. Agent mode interactions, Coding Agent tasks, and advanced model queries draw from your plan's included credits, and you are charged $0.01 per additional AI credit if you exceed the allotment.
Try This
If you don't have a Copilot plan yet, visit github.com/features/copilot/plans and activate the Free tier on your GitHub account. Then open VS Code and install the GitHub Copilot extension from the marketplace. We will configure it fully in the next lesson, but having it installed puts you one step ahead.
- GitHub Copilot has three interaction modes: inline completions, Copilot Chat (Ask/Edit/Agent), and the async Coding Agent — each solves a different kind of task
- Copilot supports multiple models including GPT-4o, GPT-4.1, Claude Sonnet/Opus, and Gemini — you choose per query in Chat
- Code completions do not consume AI credits; agentic interactions draw from your plan's included credit allotment at $0.01 per additional credit