GitHub Copilot Max costs $100 a month and includes 20,000 AI credits, which is $200 of model usage at GitHub's published rate of one cent per credit. It sits above Pro+ ($39, 7,000 credits) and Pro ($10, 1,500 credits) as the largest individual plan GitHub sells. It is worth buying if you run agent work most of the working day and you are already paying overage on Pro+; it is not worth buying for autocomplete, because code completions are unlimited and free on every paid plan including the $10 one.
Max arrived alongside a quieter change that matters more than the new tier itself: GitHub split every plan's allowance into base credits and a flex allotment. Almost every write-up of Copilot Max reports the headline credit number and stops there. The split is where the real terms are, and it changes how you should think about the value of all three individual plans.
How much does GitHub Copilot Max cost?
$100 per user per month. That is the sticker price and, unlike the plans below it, there is no tier above it on the individual side. If you need more than Max provides, the next step is not another plan, it is usage-based billing on top of this one, or a move to an organization plan.
For context, here is the full individual ladder as GitHub publishes it:
| Plan | Price | Base credits | Flex allotment | Total monthly credits | Usage value |
|---|---|---|---|---|---|
| Free | $0 | None | None | 2,000 completions, 50 chat requests | n/a |
| Pro | $10 | 1,000 | 500 | 1,500 | $15 |
| Pro+ | $39 | 3,900 | 3,100 | 7,000 | $70 |
| Max | $100 | 10,000 | 10,000 | 20,000 | $200 |
Look down the base credits column and a pattern appears immediately. $10 buys 1,000 base credits. $39 buys 3,900. $100 buys 10,000. In every case the base allowance is exactly one credit per cent of subscription, a clean one to one match with what you pay. That is not a coincidence, and GitHub says so.
How many credits does GitHub Copilot Max include?
20,000 a month: 10,000 base credits plus a 10,000 credit flex allotment. The distinction between those two halves is the single most useful thing to understand about Copilot pricing right now, and it is buried in the docs rather than the pricing page.
Base credits are matched one to one with your subscription price and never change. The flex allotment is the variable part, and GitHub states directly that flex allotments will vary over time as model pricing, new models and efficiency improvements change the underlying economics. Your base credits get spent first; once they run out, the flex allotment kicks in automatically at the same rates across the IDE, github.com and the Copilot CLI.
So on Max, half your allowance is contractual and half is discretionary. Today the plan returns $200 of usage for $100, a 2.00x ratio. The guaranteed floor of that is 1.00x. This is not a reason to avoid the plan, and it is not GitHub being sneaky, since they published the mechanism plainly. It is a reason to budget against the base number and treat the flex as a bonus that is being tuned rather than a permanent term of the deal.
GitHub Copilot Max vs Pro+: which should you buy?
Pro+ is $39 for 7,000 credits. Max is $100 for 20,000. You are paying 2.6 times as much for 2.9 times the credits, so Max is modestly better value per credit, and the gap is smaller than the price jump makes it feel.
The honest test is not which is better value in the abstract, it is whether you actually exhaust 7,000 credits. Most developers do not. Completions never touch the meter, so the only things spending it are chat, agent mode, code review and the CLI. If you use the agent a few times a day and chat when you are stuck, Pro+ is generous and Max is money set on fire. If you run multi-file agent tasks continuously, review pull requests with Copilot, and reach for the CLI as a default, 7,000 goes faster than you would guess and the overage on Pro+ will pass $61 a month quickly, which is the exact point Max starts paying for itself.
The clean decision rule: look at last month's overage on Pro+. If it exceeded $61, Max is cheaper. If it did not, stay where you are. That is the whole calculation, and it needs no forecasting, because your own usage history already answered it.
GitHub Copilot Max vs Claude Max: are they the same thing?
They share a word and nothing else. Anthropic's Claude Max plans are $100 and $200 a month and govern access to Claude, including Claude Code, and they meter you in time rather than money: a rolling five hour window plus a weekly window, shared across Claude chat and Claude Code. When you exhaust a window you wait for it to reset. Your bill does not move.
GitHub Copilot Max is $100 and meters you in credits. When you exhaust them, you either set a dollar budget and keep going, or premium features pause until the month resets. The two products fail in opposite directions: Copilot can cost more than you planned, Claude can stop when you need it most. Neither is strictly better and the choice comes down to which of those you can tolerate. We work through that trade in detail in Cursor vs Claude Code.
What counts as a request on GitHub Copilot Max?
Credits are spent per premium request, and each model carries its own multiplier, so a request to a frontier reasoning model costs more credits than a request to a light one. This is why credit counts are hard to translate into a number of prompts: 20,000 credits might be 20,000 cheap requests or a small fraction of that on the most expensive model.
What does not spend credits is worth repeating because it reverses a lot of buying decisions. Code completions and next-edit suggestions are unlimited on every paid plan and draw nothing from the meter. If autocomplete is the feature you actually value, the $10 Pro plan gives you exactly the same unlimited completions that the $100 plan does. Paying ten times more for the same autocomplete is the most common Copilot buying mistake.
Is GitHub Copilot Max worth it?
For a specific and fairly narrow group, yes. Max is worth $100 a month if agent work is most of your working day, you have measured overage above $61 on Pro+, and the alternative is being interrupted mid task. For that developer it is straightforwardly cheaper than the plan below it plus overage, and it removes a recurring decision about whether to keep going.
For most people it is not. If you are on Pro at $10 and occasionally brush the limit, Pro+ at $39 is the upgrade to make, not Max. If you have never checked your credit usage, check it before you buy anything, because the answer is usually that you are nowhere near the ceiling. And if you are buying for a team rather than yourself, ignore the individual ladder entirely: Copilot Business is $19 a seat with credits pooled across the organization, which handles uneven adoption far better than paying for individual headroom that some seats will never touch. Our GitHub Copilot pricing breakdown covers every plan, and Copilot Business seat pricing covers the team maths.
One framing that helps at this price point: $100 a month is $1,200 a year per developer, which is real budget. It is worth asking what else that buys before committing. A few hours of specialist contract help is in the same range, and for some backlogs that is the better purchase. The point is not that the tool is overpriced, it is that at $1,200 a year the spend deserves the same scrutiny as any other line item, and most teams approve dev tooling with far less thought than they give a contractor invoice.
The question Max does not answer
Every tier of Copilot, Max included, is priced around how much model usage you get. None of them changes the shape of the work: you are still in the editor, still driving, still the one who reads the diff, stages it and opens the pull request. Buying more credits buys more of the same interaction, faster. That is genuinely valuable, and it is also worth knowing what you are and are not solving with the upgrade.
Agentcode takes the other approach. You describe a task, the agent plans it, edits the code, runs your test suite, and opens a reviewable pull request on GitHub or GitLab. It never merges on its own. The price is flat at $29 a month with no credit pool and no model multipliers, so there is no overage to forecast and no tier to outgrow. It does not do autocomplete, which is why plenty of people run it alongside a $10 Copilot seat rather than instead of one. If you are weighing the wider field first, see Cursor vs Copilot or our comparison of flat and credit-metered pricing across every major tool.
All figures read from github.com/features/copilot/plans, docs.github.com and the GitHub blog in August 2026. Copilot pricing has changed twice in the past year, so check the current terms before you buy.
Keep reading
- GitHub Copilot Enterprise pricing against Copilot Pro+: which $39 seat actually buys more AI credits
- Claude Code Bedrock pricing: what an AWS team actually pays, and what the AWS bill will not tell you
- Codex VSCode extension pricing: which plan pays for Codex in VS Code, and what each one costs
- CodeRabbit pricing against GitHub Copilot code review and Greptile: what a team of ten actually pays