GitHub Copilot
The assistant that lives inside the editor, the PR and the issue tracker.
What it is
The default choice for a team already on GitHub, because the assistant is not just an editor overlay — it reviews pull requests, picks up assigned issues and runs in the cloud. The free tier is real: 2,000 completions a month is enough to evaluate it on actual work. Past that, Pro at $10 is the cheapest credible entry point among the major assistants.
Pricing updated 2026-11-11 Official pricing
Unlimited completions, cloud agent, code review, $15 monthly credits
Open pricingGitHub Copilot pricing explained
Four tiers, all per user: Free gives 2,000 completions and a limited chat allowance, Pro at $10 covers normal individual use, Pro+ at $39 raises the agent and model credits, and Max at $100 exists for people who run the agent most of the day. The thing that bites is credits rather than the subscription: heavy agent use can exhaust a month's allowance early, and the overage path is a purchase, not a throttle. Business and Enterprise add policy controls and IP indemnity, which is a procurement decision rather than a capability one. Confirm current numbers on the official plans page before budgeting.
Free-tier limits are taken from the official pricing page on the update date above. They change often — confirm before you commit to an annual plan.
GitHub Copilot key features
Inline completions and next-edit suggestions
Finishes the line you are typing and then predicts the next edit after you change something — the second part is what separates it from a plain autocomplete.
Chat and agent mode inside the editor
Ask about the file in front of you, or hand over a multi-file task and review the diff. The agent runs commands with your approval rather than silently.
Pull request review and diff explanations
Reviews a PR and explains what changed in plain language. For a reviewer who does not know the module, this shortens the first pass considerably.
Cloud agent that turns an issue into a pull request
Assign an issue and it works in a cloud environment, then opens a PR. This is the capability competitors do not have, and it only works if the code lives on GitHub.
Model selection across vendors
Claude, GPT, Gemini and others are selectable per task. You can pick a cheap model for completion and a strong one for the agent without changing tools.
Runs in almost every editor
VS Code, JetBrains, Visual Studio, Xcode, Neovim and Zed. For a team with mixed setups, that is often the deciding factor over a standalone IDE.
GitHub Copilot: how it works
Completion is a separate, latency-critical path
Inline suggestions are served by a fast model with a tight token budget, not by the chat model. That is why completion feels instant while the agent takes minutes — they are different pipelines with different cost profiles.
Repository context through indexing, not pasting
The assistant reads the repo through an index and retrieves relevant files per request. The practical limit is index freshness: work on a branch that diverges far from the default and suggestions get vaguer.
Agent execution in an isolated environment
The cloud agent runs in its own sandbox with the repo checked out, so a careless command cannot damage your machine. The trade is that it cannot touch anything not in the repository.
GitHub Copilot core strengths
The lowest credible entry price
Free tier with 2,000 completions a month, then $10 for Pro. Among the tools on this list, nothing else gives a working assistant that cheaply, which makes evaluation a non-decision.
PR review and issue-to-PR have no real equivalent
Competitors improve the editor. Copilot also touches review and triage, which is where a lot of team time actually goes.
Works in the editor you already use
No migration, no new keybindings, no second IDE to keep updated. For a team of ten with mixed setups that matters more than a marginally better model.
Model choice without switching tools
When one vendor ships a better model for a task, you change a dropdown rather than your workflow.
GitHub Copilot compared with alternatives
| Tool | GitHub Copilot | Cursor | Tabnine |
|---|---|---|---|
| Best for | Teams already hosting on GitHub who want editor, review and triage covered | Individual developers who want the most mature agent-driven editor | Organisations that cannot send source code to a model vendor |
| Pricing shape | Free tier, then $10 / $39 / $100 per user per month | Usage pools from $8 to $200 per month, third-party models billed separately | Quoted per user by sales; no public price |
| Deployment and data | SaaS; cloud agent runs in GitHub-managed environments | SaaS with local codebase indexing; no self-hosting | Self-hosted, VPC and air-gapped options with a no-training clause |
Written from the official product pages listed under Sources, not copied from another directory. Prices and limits reflect the update date above; features ship faster than reviews, so treat the list as a starting point and confirm on the official site.
Pros and cons
Strengths
- Free tier plus a $10 entry plan is the lowest credible price here
- PR review and issue-to-PR flow have no real equivalent elsewhere
- Runs in almost every editor people already use
Limitations
- Usage credits can run out mid-month on heavy agent use
- Best features assume you host code on GitHub
- Advanced security and IP indemnity need an Enterprise agreement
GitHub Copilot use cases
Everyday completion
The bread-and-butter case: finishing boilerplate, tests and repetitive refactors. It pays for itself in an afternoon of typing.
Reviewing an unfamiliar PR
Ask for an explanation of a diff before reviewing it line by line — the summary tells you where to spend attention.
Turning a triaged issue into a PR
Assign a well-scoped issue to the cloud agent and review the resulting PR. Works best when the issue already states the acceptance criteria.
Cross-language migration
Porting a module to another language or framework, with the diff reviewable file by file instead of as one unusable blob.
Onboarding to a legacy repo
Ask what a file does and where its state lives; the repo index means it answers from the actual code rather than from a guess.
