GitHub Copilot

Overview

GitHub Copilot is GitHub's first-party AI coding assistant. It goes far beyond completion with Agent Mode, Coding Agent, Spark, and Agentic Code Review. Agent Mode is generally available on both VS Code and JetBrains.

Key Features

Agent Mode

Autonomous coding agent, GA on both VS Code and JetBrains. Plans the work, edits files, and runs tests on its own.

Agentic Code Review

Code review gathers full project context and can pass suggestions directly to the coding agent for automatic fix PRs.

GitHub Spark

Pro+ and above, plus Enterprise. Describe an application in plain English to get generated code with live preview.

Semantic Code Search

Finds conceptually related code instead of keyword matches. Describe "login bug" to locate auth middleware even without the word "login."

Model Selection

The model picker offers Anthropic's Claude Opus 5 and Claude Sonnet 5 alongside OpenAI's GPT-5.6 family. Claude Opus 5 is available on Pro+, Max, Business, and Enterprise; Business and Enterprise admins must enable the policy in Copilot settings first.

Custom Endpoints

Point Copilot at your own backend URL or self-hosted model so the Copilot UI acts as a front end for internal inference.

Pricing

On June 1, 2026, Copilot moved from premium-request quotas to usage-based billing with "GitHub AI Credits" tied to token consumption. Each plan bundles a monthly credit allowance, and paid plans can purchase more beyond it.

Plan	Monthly	Included AI Credits	Notes
Free	$0	Limited	2,000 completions + 50 chats/mo
Pro	$10	$15	Unlimited completions
Pro+	$39	$70	Spark, top-tier models
Max	$100	$200	Largest allowance
Business	$19/user	$19	Org-pooled, team management
Enterprise	$39/user	Org-pooled	SSO, audit logs

Code completions and inline suggestions are unlimited on paid plans and consume no credits at all. Only chat, Agent Mode, code review, Copilot cloud agent, Copilot CLI, and Copilot Apps draw from the credit pool.

On August 28, 2026, GitHub announced Business/Enterprise seats move to prepaid billing: starting with the October 1, 2026 billing cycle, every assigned seat is charged upfront with no prorated refund on revocation. List prices are unchanged.

Typical Uses

• Inline code completion in the IDE
• Autonomous implementation via Agent Mode
• Automated PR review with Agentic Code Review
• Generating apps from natural language with Spark
• Understanding a codebase via Semantic Code Search

Strengths

• ✅ Broad IDE support (VS Code, JetBrains, Xcode, etc.)
• ✅ Agent Mode GA on JetBrains
• ✅ Agentic Code Review for automatic fix PRs
• ✅ Seamless GitHub ecosystem integration
• ✅ Claude Opus 5 / Sonnet 5 and GPT-5.6 model selection
• ✅ Unlimited, credit-free code completion on paid plans
• ✅ Personal free plan available

Weaknesses

• ❌ Token-metered billing makes heavy agent use hard to budget
• ❌ Spark limited to Pro+ ($39/mo) and above
• ❌ Less autonomous than Cursor or Claude Code
• ❌ From October 2026, Business/Enterprise seats are prepaid with no prorated refund when a seat is revoked

Official Resources

• Website: https://github.com/features/copilot
• Pricing: https://github.com/features/copilot/plans
• Docs: https://docs.github.com/en/copilot