GitHub Copilot became generally available to individuals on June 21. The launch turned a free preview into a subscription decision, while questions about suggestions resembling public code remained unresolved. I covered what a year of Copilot use actually felt like separately; this is about who could buy it, whether the price makes sense, and what to check before using suggestions in commercial code.
Who Could Buy Copilot at Launch?#
GitHub’s GA announcement offered an individual plan for $10 per month or $100 per year, free for verified students and maintainers of popular open-source projects. This was not a general free tier. The annual option saved $20 compared with paying monthly for twelve months, but only if you expected to use it for the year.
The distinction for organizations matters: GitHub said it would begin offering Copilot to companies later in 2022. An individual subscription in June was not an enterprise rollout with centralized policy and billing. Teams weighing access for employees should separate the price an individual sees from the administrative controls an organization needs.
Is $10 a Month Worth It?#
A paid seat is worth evaluating against work actually completed, not the number of suggestions accepted. Try a short period on repetitive tasks such as test scaffolding and API boilerplate; count review and rework time alongside any typing saved. A completion that passes a quick glance but introduces a subtle bug can erase the apparent gain. Treat the subscription price as an experiment, not a productivity guarantee.
For someone who uses it every day, $10 may be an easy experiment. For a team, multiply the seat count and consider onboarding, code review, and legal policy work before assuming every developer needs a license.
Public-Code Matches Need a Separate Check#
Copilot was trained on publicly available code, so teams shipping proprietary software should have a policy for suggestions that resemble licensed source. The open-source licensing debate during Copilot’s preview explains why the launch did not settle that question. GitHub later clarified that it introduced a setting in June 2022 to block suggestions of 150 or more characters matching public code, ignoring whitespace. That is a narrow matching filter, not a determination that shorter or altered suggestions are free of licensing concerns. Nor does enabling it replace reviewing the code you ship.
The question of how training and generated output interact with open-source licenses was unresolved at launch. For a commercial codebase, flag a suspiciously familiar block for review, check its provenance where possible, and get legal advice for decisions that depend on a license interpretation rather than assuming a filter settles it.
A Buying Checklist for Teams#
- Confirm eligibility: Is this an individual paid seat or a verified student/open-source-maintainer entitlement? Do not assume the later company offering was part of the June GA launch.
- Measure the workflow: Compare accepted, reviewed code and time spent fixing mistakes, not just completion counts.
- Set a code-review policy: Decide how to handle public-code matches and security-sensitive suggestions before rolling the tool out across a team. Check current product settings rather than relying on a 2022 snapshot.
GitHub’s later Copilot for Business announcement shows how organization-level purchasing and policy became a separate product decision after the individual GA launch.
My Take#
I’d trial the individual plan before committing a team to paid seats. The price is clear, but the useful output per dollar depends on what your developers actually build and how carefully they review suggestions. The free eligibility rules and the not-yet-launched company offering matter more to a purchasing decision than any demo of Copilot completing a function.
The licensing question is a separate gate. Matching filters can reduce one obvious risk, but they cannot tell you whether a particular suggestion is appropriate to include in your product. Establish review rules first; let the trial answer the productivity question, not the legal one.
