GitHub Copilot
## what it does
Overview
GitHub Copilot is Microsoft/GitHub’s AI pair programmer, integrated into VS Code, Visual Studio, JetBrains, Xcode, Neovim, Eclipse, Raycast, the GitHub CLI, and the GitHub Copilot app. The product is now an agent platform — cloud agents, Copilot code review, third-party agents (Claude Code, Codex) and MCP server controls — metered by GitHub AI Credits (1 credit = $0.01 USD).
Key Strengths
| Capability | Rating | Notes |
|---|---|---|
| Model breadth | ⭐⭐⭐⭐⭐ | Claude Haiku 4.5, Claude Opus 4.7/4.8/5, Claude Sonnet 4.6/5, Claude Fable 5/5.1, Gemini 3.5/3.6/3.7/3.8 Flash, GPT-5 mini, GPT-5.3-Codex, GPT-5.4 — all on paid tiers |
| Inline completions | ⭐⭐⭐⭐ | Unlimited on every paid plan; capped at 2,000/mo on Free |
| Cloud agents + code review | ⭐⭐⭐⭐⭐ | Background agents drive PRs; review on every PR even without a seat |
| Enterprise readiness | ⭐⭐⭐⭐⭐ | SSO, audit logs, IP indemnity (Business+), network allow-listing, MCP allow-listing, SCIM |
| IDE coverage | ⭐⭐⭐⭐⭐ | VS Code, Visual Studio, JetBrains, Xcode, Neovim, Eclipse, Raycast, GitHub CLI |
Plans (fetched 2026-09-10)
Sources: github.com/features/copilot and docs.github.com/copilot/get-started/plans. Pro, Pro+, Max are individual; Business and Enterprise are seats. All paid plans include the Copilot app and Copilot CLI. Verified students get the GitHub Copilot Student plan for free.
| Plan | USD / month | AI credits / mo | Models |
|---|---|---|---|
| Free | $0 | Allowance | Auto only (Haiku 4.5, GPT-5 mini) |
| Pro | $10 | 1,500 total (1,000 base + 500 flex) | A selection — Claude Sonnet 4.6/5, Haiku 4.5, Gemini 3.5/3.6/3.7/3.8 Flash, GPT-5 mini, GPT-5.3-Codex, GPT-5.4 |
| Pro+ | $39 | 7,000 total (3,900 base + 3,100 flex) | Premium — adds Claude Opus 4.7/4.8/5, Claude Opus 4.8 (fast mode preview), Claude Fable 5/5.1 |
| Max | $100 | 20,000 total (10,000 base + 10,000 flex) | Priority access to premium models; new models first |
| Business | $19/seat | 1,900 total per user | Premium; access control + budget control + IP indemnity |
| Enterprise | $39/seat | 3,900 total per user | Priority access + 2× Business included usage |
Models currently in the docs table (Pro+ and up): Claude Haiku 4.5, Claude Opus 4.7/4.8/5, Claude Sonnet 4.6/5, Claude Fable 5, Claude Fable 5.1, Gemini 3.5/3.6/3.7/3.8 Flash, GPT-5 mini, GPT-5.3-Codex, GPT-5.4. Codespaces + Pull Request reviewers ship the same router.
How AI credits work
From the Copilot FAQ (fetched 2026-09-10):
- 1 credit = $0.01 USD. Credits cover chats, agents, CLI, Spaces.
- Code completions and next-edit suggestions stay unlimited on every paid plan.
- Lightweight model calls cost a fraction of a credit; long agent sessions on frontier models cost more.
- When you hit the monthly allowance you can wait, set a paid-usage budget, or switch to a cheaper model.
- Business / Enterprise admins control whether paid overage is allowed.
- Code review workflows on PRs without a Copilot license bill the organization as AI Credits (off by default; needs two policies enabled). From 1 Jun 2026 code review workflows also consume GitHub Actions minutes.
Third-party agents + MCP
Pro and up include third-party agent slots — Claude Code and OpenAI Codex are listed. Business and Enterprise gain MCP server allow-lists so admins decide which MCP servers developers can reach from their IDEs.
Compliance
- SOC 2, ISO 27001 certified (Copilot Trust Center)
- IP indemnity on Business and Enterprise
- Data not used to train on Business/Enterprise. Individual subscribers can opt out.
- Retention: prompts/suggestions retained 28 days for chat on github.com / mobile / CLI (needed for thread history); not retained for IDE chat/completions; engagement data kept 2 years; feedback kept as long as needed.
Comparisons
- Cursor vs Copilot — The defining comparison for AI coding
- AI Coding Tools Compared
- Best models by lab
Related
- Copilot Trust Center — Security, privacy, responsible AI
- Copilot changelog — Previews + weekly delta
## embed this badge
