GitHub Copilot is GitHub’s AI assistant for software work. It started as code completion in the editor and now covers chat, agents, code review and a command line tool, across several editors and on github.com. In 2026 the way it is billed changed more than the product did, which is what many buyers need to understand first.
This review explains what GitHub documents about Copilot’s features, how the move to usage-based billing works, what changed for personal data in March 2026, and who should or should not consider it. It is meant for developers, team leads and managers who need to budget for it.
How we researched this: we did not run hands-on tests for this article. It is based on the vendor’s own documentation and public sources, checked on 11 November 2026. Features and prices change, so confirm details on the vendor’s site before you buy.
Key takeaways
- GitHub moved Copilot to usage-based billing on 1 June 2026. Premium request units were replaced by GitHub AI Credits, which are consumed by token usage at each model’s published API rate.
- Individual plans listed on GitHub’s plans page are Free ($0), Pro ($10), Pro+ ($39) and Max ($100). Each paid plan includes base credits equal to its price plus a flex allotment that GitHub says may change over time.
- Code completions and next edit suggestions are described as included and unlimited on paid plans, and they do not use credits.
- Since 24 April 2026, interaction data from Copilot Free, Pro and Pro+ may be used to train GitHub’s models unless the user opts out. Business and Enterprise are excluded.
- Copilot works in several IDEs, the CLI and github.com, but the feature matrix is in public preview and support differs by editor.
What it is
GitHub’s documentation lists four Copilot surfaces: the GitHub Copilot app, Copilot CLI, Copilot on github.com, and Copilot in IDEs. In the IDEs, GitHub’s feature matrix lists chat, code completion, code referencing, workspace indexing, agent mode, edit mode, custom agents, agent skills, MCP support, custom instructions, prompt files, code review, next edit suggestions, checkpoints, vision and bring-your-own-key.
The matrix is marked as a public preview and subject to change, and support varies by editor. Visual Studio Code and Visual Studio have the broadest agent support, while Neovim supports only code completion. If you work in a less common editor, check the table before you commit.
Models
Copilot offers a selection of models from several vendors. GitHub’s model hosting page says OpenAI models are hosted by OpenAI and GitHub’s Azure infrastructure, Anthropic models by Amazon Web Services, Anthropic and Google Cloud, Google models on Google Cloud, xAI models by xAI, and some other models by Microsoft or Fireworks AI. For background on what these models are, see what is a large language model.
Who it is for
- Developers who already use GitHub and a mainstream editor and want assistance without changing tools.
- Teams on GitHub who want code review and agents inside the platform where pull requests already live.
- Organisations that need admin budgets, pooled usage and policy controls, which GitHub documents for Business and Enterprise.
- Students and some maintainers, because the docs list Student as free for verified students and Pro as free for some verified teachers and maintainers.
If you do not write code, Copilot is not aimed at you. For general-purpose assistants, see our ChatGPT review or Gemini review.
Key features (documented)
| Plan | Features GitHub lists |
|---|---|
| Free | Up to 2,000 completions a month, limited chat and agent use, Copilot CLI, community support |
| Pro | Everything in Free, plus Copilot in the cloud, code review, unlimited code completion and next edit suggestions, third-party agents (Claude Code and Codex), and model selection |
| Pro+ | Everything in Pro, plus premium models including Opus, audit logs, and what GitHub calls 4x or more included usage than Pro |
| Max | Everything in Pro+, plus priority access to new models and features and 2.9x or more included usage than Pro+ |
Code review and agents
Copilot code review is listed as available in several IDEs and on github.com. GitHub’s April announcement says code review also consumes GitHub Actions minutes, billed at standard per-minute rates, so a team using it heavily should check Actions usage as well as credits. Delegating tasks to third-party agents is marked as a preview feature on the plans page. For an explanation of agent-style tools, see AI agents explained for non-technical readers.
Pricing model
Checked on 11 November 2026, US dollars. The plans page says flex allotments may change over time, so confirm current numbers on GitHub’s site.
| Plan | Price per month | Base credits | Flex allotment | Total included |
|---|---|---|---|---|
| Free | $0 | Limited chat and agent usage | Not included | n/a |
| Pro | $10 | $10 | $5 | $15 |
| Pro+ | $39 | $39 | $31 | $70 |
| Max | $100 | $100 | $100 | $200 |
GitHub says one AI credit equals $0.01. Its documentation expresses the same allowances as 1,500 credits for Pro, 7,000 for Pro+ and 20,000 for Max. Extra credits can be purchased on Pro, Pro+ and Max but not on Free.
Timeline of the billing change
| Date | What GitHub announced |
|---|---|
| 25 March 2026 | Interaction data policy update for Free, Pro and Pro+ (effective 24 April) |
| 27 April 2026 | Announcement that Copilot moves to usage-based billing on 1 June 2026 |
| 12 May 2026 | Flex allotments added to Pro and Pro+, and a new Max plan, effective 1 June |
| 1 June 2026 | Usage-based billing begins for all plans |
Business and Enterprise
GitHub’s April announcement listed Business at $19 per user per month and Enterprise at $39 per user per month, each with matching included credits, and said included usage can be pooled across an organisation. Admins can set budgets at enterprise, cost centre and user levels, and when the pool runs out, organisations can allow extra usage at published rates or cap spending. The current plans page did not list Business or Enterprise prices and pointed to sales, so confirm those figures directly. For help with the general method, see how to read an AI tool’s pricing page.
Strengths
- Fits existing workflow. It is documented across multiple IDEs, the command line and github.com.
- Completions stay predictable. Code completions and next edit suggestions are stated not to use credits on paid plans.
- Cost controls for organisations. Budgets at several levels and pooled usage are documented.
- Several model vendors. Model choice is part of the paid plans.
- A free tier. It lets you try completions and limited chat, though with caps.
Limits and cautions
- Usage-based billing means variable value. Longer agent runs consume more credits. GitHub itself said, in its May announcement, that it adjusted individual plans after feedback that included usage might not cover longer agent runs and multi-step work.
- No fallback to a cheaper model. GitHub says the lower-cost model fallback ended on 1 June, so running out of credits has consequences.
- Flex is not guaranteed. The flex part of the allowance can vary as GitHub adjusts for model pricing and efficiency.
- Page inconsistencies. On the plans page we read, model lists looked identical across plans even though the descriptions say Pro+ adds premium models. Treat the descriptions and the billing docs together and ask GitHub if it matters to you.
- Preview features. The feature matrix and some agent features are labelled preview.
- Free is for personal use. GitHub’s docs say Free lacks access management, audit logs, policy management, file exclusion, usage data and indemnification, so it is not suited to enterprises.
As with any AI coding tool, suggested code can be wrong or insecure. Treat output as a draft and see AI hallucinations: what they are and how to reduce them.
Privacy and data
This is the area where individual and business plans differ most. GitHub announced on 25 March 2026 that, from 24 April, interaction data from Copilot Free, Pro and Pro+ users would be used to train and improve its models by default unless users opt out.
| Topic | Individual plans (Free, Pro, Pro+) | Business and Enterprise |
|---|---|---|
| Interaction data for training | Used by default unless the user opts out | Not affected by the programme |
| Data included | Accepted or modified outputs, inputs, code context, comments, file names, repository structure, feature interactions, thumbs up and down feedback | Not applicable |
| Opt-out | Copilot settings under Privacy at github.com/settings/copilot | Not applicable |
| Sharing | May be shared with GitHub affiliates including Microsoft, not with third-party AI model providers | Not applicable |
GitHub says content from issues, discussions or private repositories at rest is excluded, but code from private repositories that Copilot processes during active use can be used unless the user opts out. People who previously opted out keep that choice. On model providers, GitHub’s hosting page describes zero data retention or no-training arrangements for several vendors, with exceptions such as certain Claude models that Anthropic retains for safety classifiers. Read our guide to protecting company data when using AI tools before using any individual plan for client code.
Alternatives
- Cursor if you want an agent-first editor with its own plan structure. See the Cursor review.
- Claude with Claude Code if you prefer a vendor that sells its own models. See the Claude review.
- A general chatbot for occasional snippets, if you do not need editor integration.
Who should and should not use it
Best for you if
- Your team lives on GitHub and wants AI in the editor, CLI and pull requests under one vendor.
- You want budget controls and pooled usage, and will review spend regularly.
- You are on Business or Enterprise, where the interaction-data training programme does not apply.
Look elsewhere or take care if
- You use an individual plan for confidential code and have not opted out of data use.
- You need a fixed bill and expect heavy agent use.
- Your editor is poorly supported in the feature matrix.
Questions to ask before buying
- Which editor features do I need, and does the matrix support them in my IDE?
- How many credits will my team realistically use, and who sets budgets?
- Have all users reviewed the Privacy setting and chosen to opt out or not?
- Do I need Business or Enterprise for policy, audit and indemnification?
- Will code review also use Actions minutes in our setup?
- What happens to my annual plan if I bought before 1 June 2026? GitHub’s announcement says annual subscribers keep premium request pricing until the plan expires.
Trial it on real work with how to test an AI tool before you pay for it.
Frequently asked questions
Is GitHub Copilot free?
There is a Free plan with up to 2,000 completions a month and limited chat and agent use, according to GitHub. Verified students and some teachers and maintainers may qualify for paid features free of charge.
What are GitHub AI Credits?
The unit that replaced premium request units on 1 June 2026. GitHub says one credit equals $0.01 and credits are consumed by token usage at each model’s published API rates.
Do code completions use credits?
GitHub says code completions and next edit suggestions stay included and do not use credits on paid plans.
Does Copilot train on my code?
For Free, Pro and Pro+, GitHub says interaction data may be used for training unless you opt out. Business and Enterprise are excluded.
What is the Max plan?
A $100 per month individual plan announced on 12 May 2026, with $100 base credits and $100 flex allotment.
How do I choose between coding assistants?
Use how to compare AI tools side by side fairly with a task from your own codebase.
Conclusion
Copilot’s documented strengths are reach across editors and GitHub, plus predictable treatment of completions. Its main cautions are usage-based billing for chat and agents and the default data-use policy on individual plans. Check both before adopting.
Start with how to choose an AI tool, check the vendor’s current plans page, and decide your data settings before anyone pastes in client code.
Sources
- GitHub Copilot plans, accessed 11 November 2026
- Plans for GitHub Copilot individuals (GitHub Docs), accessed 11 November 2026
- GitHub Copilot is moving to usage-based billing (GitHub Changelog), 27 April 2026, accessed 11 November 2026
- Copilot individual plans: flex allotments and a new Max plan (GitHub Blog), 12 May 2026, accessed 11 November 2026
- Updates to GitHub Copilot interaction data usage policy (GitHub Blog), 25 March 2026, accessed 11 November 2026
- Copilot feature matrix (GitHub Docs), accessed 11 November 2026
- Hosting of models for GitHub Copilot (GitHub Docs), accessed 11 November 2026
