GitHub Copilot Features
This page provides a comprehensive overview of GitHub Copilot plans, combining official features with example videos. For the most current pricing, visit GitHub's official pricing page. For the most current pricing, visit GitHub's official pricing page. For the most recent plan details, view the official documentation.
Starting June 1, 2026, GitHub Copilot is switching from request-based to usage-based (token-based) billing.
2026
Token-based billing: from premium request units to AI credits and tokens
Replaces fixed monthly premium request quotas with a flexible token-based billing model. Cost scales with actual model usage — heavier models like Claude Opus consume more tokens per interaction while lighter models cost less. Gives individuals and organizations fine-grained control over AI spend without hard request caps.
Organization-wide Custom Instructions
Lets administrators define Copilot instructions that apply across all repositories in an organization. Enforce coding standards, preferred libraries, security practices, and domain context centrally — every developer's Copilot interactions automatically follow these guidelines without per-repo setup.
2025
GitHub Spark
Describe an app idea in natural language and Spark scaffolds, deploys, and hosts a working web app in minutes. Includes built-in data storage, AI capabilities, and one-click sharing. Aimed at turning ideas into shipped micro-apps without manual infrastructure setup or coding knowledge.
Model selection
Lets you choose which AI model powers Copilot Chat — switch between OpenAI, Anthropic, Google, and other frontier models for different tasks. Use a faster model for quick questions and a more capable one for complex reasoning, all without leaving the editor.
Bring Your Own LLM
Supply your own API key for Anthropic, OpenAI, Google, or any OpenAI-compatible provider inside VS Code Copilot Chat. Routes requests through your own contracts and quotas while keeping the full Copilot UX, and unlocks models not yet in the default catalog.
Copilot Coding Agent in Pull Requests
Assign a GitHub Issue to Copilot and it works autonomously in the background — creating a branch, reading related code, implementing a solution, running CI checks, and opening a pull request for your review. Reduces context-switching by handling implementation work while you focus on other things.
MCP with Azure and GitHub
Connect MCP (Model Context Protocol) servers to Copilot Agent Mode to give agents access to live external data and actions — query Azure resources, create GitHub issues, read from databases, and more. Agents use the results directly in their reasoning, enabling real-world integrations beyond the local codebase.
Agent Mode
Copilot independently plans and executes multi-step changes from a single natural-language prompt — edits files across the repository, runs terminal commands, executes tests, reads error output, and self-corrects. You review each proposed change before it is applied, staying in control throughout.
Premium Requests
A monthly allowance of requests that unlock the most capable frontier models such as Claude Sonnet 4, GPT-5, and Gemini 2.5 Pro. Lighter models remain unlimited while top-tier ones draw from this pool, giving teams predictable access to high-capability AI without unbounded cost.
Next Edit Suggestions
Rather than only completing the current line, Copilot predicts the next logical edit anywhere in the open file based on your recent changes. Particularly useful for repetitive refactors, renames, and pattern-following changes that would otherwise require manual fan-out across multiple locations.
GPT-4o Copilot Suggestions Model
Upgrades the underlying model for inline code completions from GPT-3.5 Turbo to GPT-4o, resulting in more accurate, context-aware suggestions. Better at understanding complex codebases, handling multi-line completions, and producing idiomatic code in a wider range of languages.
Fetch Webpage
Include the content of any public web page as part of your Copilot Chat context. Useful for asking Copilot to implement something based on documentation, a spec, or an API reference — paste the URL and Copilot retrieves and reads the page as part of your prompt.
Set Coding Guidelines for Code Review
Define custom coding standards that Copilot applies when reviewing pull requests. Specify style rules, security requirements, preferred patterns, and naming conventions — Copilot enforces them consistently across all reviews without relying on individual reviewers to catch them manually.
Copilot Spaces
Create purpose-built AI agents enriched with specific repositories, files, and guidelines. Spaces let you scope Copilot's knowledge to a particular project, team, or domain — resulting in more relevant answers and suggestions that reflect your organizational context.
Ice Breakers
Generates project starter templates and boilerplate code to accelerate the beginning of new work. Instead of setting up structure from scratch, describe what you want to build and Copilot produces a ready-to-use starting point tailored to your stack and requirements.
Code Completion
Real-time AI code suggestions as you type — from single-line completions to entire function bodies. Uses your open files, comments, and cursor context to deliver language-aware completions across all major editors and languages. The foundational Copilot capability available since launch.
Content Exclusions
Prevents specific files, directories, or patterns from being sent to Copilot's AI models. Useful for excluding files containing secrets, proprietary algorithms, or sensitive data you do not want included as context in completions or chat responses.
Usage Metrics
Provides visibility into how Copilot is being used across your team or organization — acceptance rates, active users, most-used features, and more. Helps managers and admins understand ROI, identify adoption gaps, and make informed decisions about Copilot rollouts.
Chat in IDE
A conversational AI assistant embedded directly in your IDE. Ask questions about your code, get explanations, generate tests, fix bugs, and refactor through natural language without leaving your editor. Supports multi-turn conversations with full awareness of your open files and workspace.
Unit Tests
Automatically generates unit tests for your code — covering happy paths, edge cases, and error conditions. Reduces the time spent writing boilerplate test scaffolding and helps ensure test coverage for functions and classes you have already written.
Commit Message Custom Instructions
Configure how Copilot generates commit messages by providing custom instructions — specify preferred format, required prefixes such as Conventional Commits, character limits, or project-specific conventions. Ensures generated messages consistently match your team's standards.
Playwright Test Generation
Generates Playwright end-to-end tests from your existing application code. Copilot analyzes page structure and interactions to produce test scripts that cover key user flows, reducing the manual effort of writing browser automation from scratch.
2024
Code Translation
Converts code from one programming language to another while preserving logic and intent. Useful for migrating legacy codebases, porting scripts between ecosystems, or learning how a pattern works in a different language.
@github Chat Participant
The @github chat participant brings repository-aware context into Copilot Chat on GitHub.com — ask about issues, pull requests, commits, code, and discussions across your repositories. Answers are grounded in your actual repo data rather than generic knowledge.
Code Docs
Generates documentation for your code — inline comments, JSDoc/XML doc blocks, README sections, and more. Explains what functions do, their parameters, return values, and side effects, reducing the documentation debt that builds up during fast development cycles.
Chat on Mobile
Brings Copilot Chat to the GitHub mobile app, letting you interact with your repositories on the go. Summarize issues, pull requests, and discussions, ask questions about a codebase, and get AI-assisted insights without needing to be at a computer.
Commit Message Suggestions
Analyzes your staged changes and proposes an appropriate commit message that summarizes what was changed and why. Saves time writing messages manually and produces more consistent, descriptive commit history.
Multi-file Edits
Describe a change in natural language and Copilot proposes coordinated edits across all relevant files simultaneously. You review each change as a diff per file, making it practical to execute refactors, renames, and feature additions that touch many parts of a codebase at once.
PR Body Generation in WebUI
Automatically generates a pull request description from your diff and branch context on GitHub.com. Produces a summary of what changed and why, saving time writing PR bodies and ensuring reviewers have the context they need.
Local In-Editor Code Reviews
Review code changes directly within your IDE before committing. Copilot highlights potential issues, suggests improvements, and provides inline feedback on your local diff — bringing AI code review into your normal editing workflow rather than only on pull requests.
Multi-Models
Choose from multiple frontier AI models within Copilot Chat — including models from OpenAI, Anthropic, and Google. Different models have different strengths, letting you pick the best one for your specific task without switching tools.
Copilot Extensions Marketplace
A marketplace of third-party extensions that plug directly into Copilot Chat. Lets you query, configure, and trigger external developer tools — databases, observability platforms, feature flags, CI systems, and more — through natural language without leaving your editor.
Fine Tuned Models
AI models trained specifically on your organization's private codebase to produce suggestions aligned with your internal patterns, APIs, and conventions. Results in higher-relevance completions than general-purpose models when working within your own code.
Path-specific Custom Instructions
Define different Copilot instructions for different parts of your codebase — for example, stricter security rules for authentication code or React conventions only for frontend files. Instructions are applied automatically based on which files are involved in the current context.
PR Body Generation in VS Code
Generates pull request descriptions directly from VS Code before you push. Analyzes your local branch changes and produces a structured summary, so you arrive at GitHub with a completed PR body rather than starting from scratch.
User Instructions
Lets you write personal instructions that persistently guide Copilot's behavior in VS Code — specify your preferred language version, coding style, libraries to avoid, or project context. Applied automatically to every chat interaction without repeating yourself each session.
Code Explanation
Explains what a piece of code does in plain language — covers logic flow, algorithm intent, data transformations, and edge cases. Useful for understanding unfamiliar code, onboarding to a new codebase, or reviewing legacy implementations.
CLI Assistance
Provides natural-language assistance for the command line via gh copilot. Use suggest to generate shell commands from a description, and explain to understand what a complex command does. Covers git, the GitHub CLI, and common shell tools — and can execute the generated command after your confirmation.
Security Advisory Summaries
Generates plain-language summaries of Dependabot security alerts directly in GitHub. Explains the vulnerability, affected versions, and recommended remediation steps, making it faster to understand and act on security issues without consulting external documentation.
Preview for Copilot Workspace
A cloud-based environment where you describe a task and Copilot proposes a multi-file implementation plan, generates the changes, and runs tests without a local setup. An early exploration of plan-driven, agentic development that informed later features like the Coding Agent.
Copilot Metrics API
A REST API that exposes Copilot usage data programmatically — acceptance rates, active users, model usage, seat counts, and more. Lets teams integrate Copilot metrics into dashboards, reporting pipelines, and adoption tracking tools of their choice.
Code Scanning AI Autofix
When Code Scanning detects a security vulnerability, Copilot automatically generates a code fix and presents it as a suggested change in the pull request. Reduces the time between detection and remediation by making fixes immediately actionable without leaving the review flow.
Web Search with Bing
Enables Copilot to search the web for current information and include it in responses. Useful for questions about recent library updates, unfamiliar APIs, or anything that requires up-to-date context beyond the training data.
Chat with Knowledge Bases
Lets Copilot answer questions grounded in your organization's private repositories and wikis, indexed as Knowledge Bases. Results in contextually relevant answers about your internal codebase, architecture decisions, and domain-specific conventions rather than generic responses.
Chat with your Pull Request
Ask Copilot about the code in a specific pull request — get summaries, explanations of individual changes, suggestions for improvements, or answers about the intent behind the diff. Available directly on the PR page in GitHub.
Multi-Repository Context
Allows Copilot to draw context from multiple repositories simultaneously in GitHub Enterprise environments. Useful for answering questions about systems that span several repos, finding cross-repo patterns, and getting suggestions that account for shared libraries or APIs.
Smart Actions
Context-sensitive AI actions surfaced inline in the editor and on GitHub.com — such as explain, fix, document, or review. Provide one-click access to the most relevant Copilot operation for the code or context currently in focus.
2023
Code Debugging
Helps diagnose bugs by analyzing the call stack, variable values, and error messages in context. Suggests what went wrong and proposes fixes, reducing the time spent mentally tracing execution paths during debugging sessions.
Code Fixing
The /fix slash command in Copilot Chat analyzes an error or problematic code selection and proposes a corrected version. Handles compilation errors, runtime exceptions, type mismatches, and logic issues — often resolving them in a single interaction.
Slash Commands
Slash commands in Copilot Chat provide quick shortcuts for common operations — /explain, /fix, /tests, /doc, and more. Reduce the need to phrase natural-language requests for repetitive tasks and provide predictable, consistent entry points into Copilot's capabilities.
Explain Failed Action Jobs
When a GitHub Actions job fails, Copilot analyzes the error logs and explains what went wrong in plain language. Suggests likely causes and remediation steps, reducing time spent digging through CI output to understand build and test failures.
User Management
Centralized controls for assigning, removing, and auditing Copilot licenses across an organization. Administrators can manage access per user, team, or all members, and review seat usage to optimize license allocation.
Integration in Web UI
Copilot is embedded throughout GitHub.com — available in the code editor, issues, pull requests, and discussions. Lets you use AI assistance without switching to a separate IDE, useful for quick reviews, issue triage, and lightweight edits directly in the browser.
Data Excluded From Training by Default
Code, prompts, and suggestions are not used to train GitHub's AI models by default for Business and Enterprise customers. Your private code remains private and does not contribute to model improvements that could surface in other users' suggestions.
2022
Code Referencing
When a Copilot suggestion closely matches code in a public GitHub repository, Copilot shows a reference indicating the source. Helps you make informed decisions about using the suggestion and provides attribution information for compliance and legal awareness.