What it is
GitHub Copilot started as autocomplete and is now a set of tools across editors, the command line, GitHub.com and the mobile app. It offers inline and next edit suggestions, chat and agent mode in VS Code, JetBrains and other IDEs, the Copilot CLI, pull request code review, and a cloud agent you can assign GitHub issues to so it opens a pull request.
Its model list is unusually broad. GitHub's supported-models page lists OpenAI (GPT-5.6 and GPT-6, including GPT-6 Astra), Anthropic (Claude Opus 5.5, Sonnet 5.5 and Fable), Google Gemini, xAI Grok, Moonshot's Kimi K3 and Microsoft's MAI-Code. Pro+ and Max can also hand tasks to third-party agents such as Claude and OpenAI Codex (in preview).
On 1 June 2026 GitHub replaced premium requests with usage-based GitHub AI Credits. One credit is $0.01, and a request costs whatever its tokens cost at the model's API rate. Completions and next edit suggestions do not use credits.
Our review
Copilot's advantage is where it lives. It is built into GitHub, so you can assign an issue to the cloud agent, get a pull request back and have Copilot review other people's pull requests without adding a vendor or an editor. For companies already on GitHub Enterprise Cloud, the policy controls and IP indemnity on Business and Enterprise are hard for startups to match. One gap: Copilot is not available for GitHub Enterprise Server.
Model choice is the other strength. GitHub's supported-models page lists GPT-6 Astra, Claude Opus 5.5 and Fable, Gemini 4 Argon, Grok 4.7, Kimi K3 and Microsoft's MAI-Code. That breadth matters more since OpenAI said it would cut off Cursor's direct access on 12 November 2026; Copilot is now one of the few multi-vendor tools with both GPT-6 and Claude.
The weak point is billing. Since 1 June 2026, chat, agents, code review and the CLI spend AI Credits at each model's token rate, and the old fallback to a cheaper model when you run out is gone. Code review on private repositories also uses GitHub Actions minutes. The individual plans carry a bonus, $15 of credits for $10 on Pro and $200 for $100 on Max, though GitHub says the flex part may change. Business gets 1,900 credits, exactly its $19 price, pooled across the organisation so light users offset heavy ones, and community threads report Pro users exhausting a month's credits in a couple of long agent prompts.
On data, the business tiers earn their price. For Business and Enterprise, IDE chat, completions and CLI inputs and outputs are not retained by default, other surfaces keep them for up to 28 days, and the data is excluded from training under GitHub's Data Protection Agreement. Both are in GitHub's SOC 2 Type 2 and ISO 27001 scope. Individual plans differ: GitHub may train on Free, Pro and Pro+ interactions unless you opt out, so a contractor on a personal Pro seat sits outside the company's controls.
Where it earns its place
A 120-developer company on GitHub Enterprise Cloud
Give every engineer completions, chat and pull request review under one contract
Business seats at $19 cost $2,280 a month and include 1,900 AI Credits each; admins set per-user budgets so a few heavy agent users cannot drain the bill. IDE prompts are not retained by default and IP indemnity applies.
A maintainer of a busy open-source library
Triage issues by handing small fixes to an agent
On Pro+ at $39 a month ($70 of credits), issues are assigned to the cloud agent, which opens a pull request for review. Agent runs bill by tokens, so setting a budget before assigning a backlog avoids a surprise overage.
A procurement manager at a regulated insurer
Check whether Copilot's data handling passes review
Business and Enterprise are covered by GitHub's DPA, SOC 2 Type 2 and ISO 27001 scope, but coding-agent session logs are kept for the life of the account and some models, such as Claude Fable 5, have provider-side retention. Self-hosted GitHub Enterprise Server is not supported.
Before you commit
- Setup
- Install the extension in VS Code, JetBrains or another supported editor and sign in with GitHub; Business requires an organisation admin to grant seats.
- Integrations
- VS Code, JetBrains and other IDEs, the Copilot CLI, GitHub.com, GitHub Mobile and MCP servers; Pro+ and Max can delegate to Claude and Codex agents in preview.
- Data and privacy
- Business and Enterprise: IDE and CLI inputs and outputs not retained by default, other surfaces up to 28 days, engagement data two years, no training; individual plans may train unless you opt out.
- Cost at scale
- 100 Enterprise seats cost $3,900 a month and include 3,900 credits per user, pooled into one 390,000-credit allowance for the organisation; usage beyond that is billed at model token rates against the budget an admin sets, and code review adds Actions minutes.
- Limits
- Free gets 2,000 completions a month and limited chat; on Business and Enterprise, if admins block extra usage, Copilot pauses until the next cycle rather than falling back to a cheaper model.
What's good
- Unlimited completions and next edit suggestions on every paid plan
- Widest model menu: GPT-6, Claude, Gemini, Grok, Kimi and Microsoft models
- Assign GitHub issues to a cloud agent and get pull requests back
- Business and Enterprise: no IDE prompt retention by default, IP indemnity, SOC 2 Type 2
- Admins can set per-user AI Credit budgets with alerts
What's not
- Token-based AI Credits make heavy agent use hard to predict
- No fallback to a cheaper model once credits run out
- Free, Pro and Pro+ interactions may be used for training unless you opt out
- Not available on GitHub Enterprise Server
- Code review also consumes GitHub Actions minutes on private repositories
Best for
Individuals who want a low-cost assistant with many models, and teams that already run on GitHub.
Alternatives
Head to head
GitHub Copilot: common questions
- Is GitHub Copilot worth it?
- The safe default for organisations already on GitHub, with the broadest model menu and real data controls, though token billing has made heavy agent use a line item to watch. We rate it 4.4 out of 5.
- How much does GitHub Copilot cost?
- Free (2,000 completions a month and limited chat); Pro $10/month with $15 of AI Credits; Pro+ $39/month with $70; Max $100/month with $200; Business $19/user/month with 1,900 credits; Enterprise $39/user/month with 3,900. 1 credit = $0.01; extra usage billed against a budget.
- What are the downsides of GitHub Copilot?
- Token-based AI Credits make heavy agent use hard to predict; no fallback to a cheaper model once credits run out; free, Pro and Pro+ interactions may be used for training unless you opt out; not available on GitHub Enterprise Server; code review also consumes GitHub Actions minutes on private repositories.
- Who is GitHub Copilot best for?
- Individuals who want a low-cost assistant with many models, and teams that already run on GitHub.
- How much does GitHub Copilot cost for a team or in production?
- 100 Enterprise seats cost $3,900 a month and include 3,900 credits per user, pooled into one 390,000-credit allowance for the organisation; usage beyond that is billed at model token rates against the budget an admin sets, and code review adds Actions minutes.
- Is GitHub Copilot safe for company data?
- Business and Enterprise: IDE and CLI inputs and outputs not retained by default, other surfaces up to 28 days, engagement data two years, no training; individual plans may train unless you opt out.
- What are the best GitHub Copilot alternatives?
- The closest alternatives we have reviewed are Cursor, Claude Code, OpenAI Codex and JetBrains AI (Junie, AI Assistant, Air).
Answers checked Oct 23, 2026.
Make GitHub Copilot? Show this rating on your site
