Cursor vs GitHub Copilot 2026: Pricing, Limits and Which to Pay For

GitHub Copilot is the cheaper tool for a developer who wants Claude or GPT models: Pro costs $10 a month and Max $100, and every plan includes a dollar-denominated credit allowance you can spend on any model in the picker. Cursor is the cheaper tool only if you are happy coding with Grok and Composer: Plus costs $8, Premium $30, Super $100 and Ultra $200, and third-party models cost extra on every plan. Prices read on both vendors' pages on October 26, 2026.

The two subscriptions now buy different things. Cursor sells a weekly pool of usage for its own models, shared with the Grok apps, and bills Claude, GPT and Gemini separately at API list price from prepaid credit. Copilot sells GitHub AI Credits, worth $0.01 each, that any supported model draws down at its per-token rate. Once you see that, the price comparison becomes a question about which models you want to run.

An AI coding assistant is a tool that reads your codebase and writes, edits, reviews or explains code from a prompt, inside an editor, a terminal or a cloud agent. Toolradar data: the AI coding ranking we publish tracks 179 AI coding tools as of October 25, 2026, and Cursor is the most cross-shopped of them, listed as an alternative to 140 other tools, with GitHub Copilot second at 129. The product records are Cursor and GitHub Copilot on Toolradar, and our roundup of the best AI coding assistants covers the wider field. Note that GitHub Copilot is a different product from Microsoft 365 Copilot, which we price on Microsoft Copilot pricing.

At a glance

Question	Cursor	GitHub Copilot	Who wins
What it is	Its own editor (a VS Code fork), a CLI and cloud agents	An extension for VS Code, JetBrains, Visual Studio, Xcode, Neovim, Eclipse and Zed, plus a CLI and app	Copilot, if you keep your editor
Cheapest paid plan	Plus, $8 a month, Grok models only	Pro, $2 a month more, any model in the picker	Cursor on price, Copilot on model choice
What the subscription includes	A weekly usage pool for Grok 4.7, 4.6, 4.5 and Composer 2.5	Monthly AI credits spendable on Claude, GPT, Gemini, Grok and others	Copilot, for flexibility
Claude and GPT models	Extra, at API list price, Premium and above only	Included in the credit allowance on every paid plan	Copilot
Allowance reset	Weekly	Monthly, on the 1st, no rollover	Cursor, for a bad week
Code completions	Included on every plan	Unlimited on every paid plan, 2,000 a month on Free	Tie
Top of the ladder	Ultra, $200 a month, 50x the Plus pool	Max, 20,000 credits a month	Depends on the models you run
Team seat	A Premium, Super or Ultra seat at the individual price, plus $0.25 per million tokens on third-party models	Business $19 per user, Enterprise twice that, pooled credits	Copilot
Training on your code	Privacy mode guarantees no training	Individual plans may train unless you opt out; Business and Enterprise do not	Cursor, by default

Both vendors rebuilt their ladders this year, so most of what you read about "Cursor Pro at $20" or Copilot's premium requests is out of date. The sections below work from the current pages only.

What each plan costs in 2026

Plan	Price	Allowance	What it adds
Cursor Free	$0	Limited Agent and Chat usage	Tab completions, no third-party models
Cursor Plus	$8 a month, or $80 a year	Base weekly pool	Grok 4.7, 4.6, 4.5 and Composer 2.5 at fixed settings, Grok Chat, Voice, Imagine and Grok API access
Cursor Premium	$30 a month, or $300 a year	4x the Plus pool	Cloud agents, Automations, Bugbot code review, Origin code hosting, third-party models at list price, Grok Bot
Cursor Super	$100 a month, or $1,000 a year	15x the Plus pool	Priority model access at peak times
Cursor Ultra	$200 a month, or $2,000 a year	50x the Plus pool	Earliest access to new products and models
Cursor Teams	The Premium, Super or Ultra price per seat a month	The matching individual pool per seat	Central billing, SSO, usage analytics, team marketplace, enforced privacy mode
Copilot Free	$0	A limited credit allowance, 2,000 completions a month	Auto model selection only, CLI and agent mode
Copilot Pro	$10 a month	1,000 base credits plus a 500 flex allotment, 1,500 in total	Unlimited completions, cloud agent, code review, MCP, a selection of models
Copilot Pro+	$39 a month	3,900 base plus 3,100 flex, 7,000 in total	Third-party coding agents (Claude, Codex) in preview, access to premium models
Copilot Max	$100 a month	10,000 base plus 10,000 flex, 20,000 in total	Priority access to premium models
Copilot Business	$19 per seat a month	1,900 credits per user, pooled	Policy controls, IP indemnity, usage metrics
Copilot Enterprise	$39 per seat a month	3,900 credits per user, pooled	Everything in Business, SAML SSO, enterprise security

Copilot's credits convert to dollars. GitHub prices every model per million tokens, in the same units as the public API prices, and converts the bill into credits at $0.01 each. Pro's 1,500 credits are $15 of model usage, Pro+ turns its price into $70 of usage, and Max doubles its price into credits, per GitHub's plans page and its billing documentation. The flex allotment is the part GitHub says it may adjust over time, so treat the base credits as the guaranteed half.

Cursor's pool has no dollar figure attached. Its pricing documentation describes Premium as 4x Plus, Super as 15x and Ultra as 50x, refreshing weekly and shared with the Grok apps, and it does not say how many tokens Plus includes. Both vendors exclude tax from the prices above, and Cursor asks you to cancel at least 24 hours before renewal.

The one difference that decides the bill

On Cursor, the subscription pays for Cursor's own models and nothing else. Claude Sonnet 5.5, Opus 5.5, GPT-5.6 Sol and Gemini are available from Premium upward, but they cost API list price from the first request, paid from a prepaid credit balance you top up yourself. Plus and Free cannot use them at all. If your daily driver is Claude, Cursor's Premium plan is the entry fee, not the whole bill.

On Copilot, the same models draw from the credits already included in the plan. Both vendors publish identical per-token rates for the models that matter: Claude Sonnet 5.5 at $2 per million input tokens and $10 per million output, Claude Opus 5.5 at $4 and $20, GPT-5.6 Sol at $4 and $20, per Cursor's model pricing and GitHub's models and pricing reference. The difference is who pays for them: on Copilot the subscription does, on Cursor you do.

Cursor's answer is Grok. Grok 4.7 costs $2 per million input tokens and $6 per million output when you buy it beyond the pool, and Composer 2.5 costs $0.50 and $2.50, so the included pool stretches a long way if you let those models do the work, and a tight prompt stretches it further (see how to write AI prompts). A developer who prefers them gets a flat monthly price with a weekly reset, which Copilot cannot match: Copilot's credits run out on the first of the month and never roll over, and a heavy fortnight leaves you buying extra usage at the same per-token rates.

The monthly cost at three usage levels

The scenarios below measure usage in dollars of frontier-model tokens at the list rates both vendors publish, because that is the unit the two ladders share. They are our arithmetic, not a vendor quote, and they assume you want Claude or GPT models rather than Grok.

Usage level	What it looks like	Cheapest Cursor route	Cheapest Copilot route
Light: completions plus a few chats a day, no third-party models	A few dollars of tokens a month	Plus (Grok only)	Pro (covers $15 of any model)
Regular: daily agent sessions, about $50 of Claude Sonnet 5.5 tokens	10M input and 3M output tokens, $20 plus $30	Premium $30 plus $50 of prepaid credit, $80	Pro+, $39 (covers $70)
Heavy: agents running most of the day, about $150 of tokens	30M input and 9M output tokens, $60 plus $90	Premium plus $150 of credit, $180	Max (covers $200)

At the regular level the gap is $41 a month, at the heavy level $80. On Cursor the right plan for a Claude user is Premium, not Super or Ultra, because the bigger pools only buy more Grok and Composer usage; the third-party bill is the same on every tier. On Copilot, Pro plus $35 of extra usage would reach $45 for the regular scenario, so Pro+ is the better buy once you pass about $15 of tokens, and Max beats Pro+ once you pass $70.

Flip the assumption and Cursor wins. A developer who codes with Grok 4.7 and Composer 2.5 pays the Premium price at both the regular and heavy levels, as long as the weekly pool holds, and the Plus price for the light level, against Copilot's Pro, Pro+ and Max prices. Grok is also in Copilot's picker at the same $2 and $6 rates, so the question is whether you want Cursor's editor and its weekly reset, or Copilot's choice of model inside the editor you already use.

A paid Copilot plan also takes 10% off model costs when you leave the model on auto selection, which stretches the credits a little further than the table shows; the table uses the undiscounted rates.

Free tiers: what you get for nothing

Cursor Free includes Agent, Chat and Tab with a limited allowance, no third-party models, and none of the cloud agents, Bugbot or Origin features. Copilot Free includes 2,000 completions a month, a limited credit allowance, agent mode, the CLI and MCP servers, with the model chosen automatically. Students get Copilot free with unlimited completions, and verified teachers and maintainers of popular open-source projects get Pro free, per GitHub's plans documentation.

Copilot Free is the more useful trial for a working developer because it runs inside your current editor and includes the agent surfaces. Cursor Free is a trial of a new editor, which is a bigger ask before you have paid anything.

Teams: Copilot prices the credits, Cursor prices the seat and the model

Five developers on Copilot Business pay $95 a month (5 seats at $19), and their 9,500 pooled credits cover $95 of model usage across the team, with admins setting a budget for anything beyond. Agent sessions are what consume those credits fastest, as our explainer on agentic AI sets out. Five Premium seats on Cursor Teams cost $150 a month, each seat keeps its own weekly pool, and third-party models are billed at list price plus a Cursor Token Rate of $0.25 per million tokens, from a prepaid balance the admin tops up. Plus seats are not available to teams.

What Cursor Teams buys for the extra money is SAML and OIDC SSO, enforced privacy mode, custom chat retention and a team marketplace for rules and skills, per its team pricing page. Copilot puts SSO on an Enterprise seat, which still comes in under a Premium seat, and Business adds IP indemnity, which Cursor does not list on any plan. Cursor's Enterprise plan has pooled usage, invoicing and SCIM on a custom quote; Copilot Enterprise requires GitHub Enterprise Cloud.

Privacy, data and lock-in

GitHub's plans page states that interaction data from individual Free, Pro, Pro+ and Max subscribers may be used to train and improve models unless you opt out in your Copilot settings, while Business and Enterprise data is not. Cursor's privacy mode, which an individual can switch on and a team admin can enforce, guarantees that code data is not used for training by Cursor or its model providers. If you are paying for yourself, Cursor protects your code by a setting, Copilot by an opt-out.

The lock-in runs the other way. Cursor is now sold as part of The X Subscription, one bill for Cursor, Grok and X managed at grok.com, and the switch for an existing customer is permanent. Copilot is a line on your GitHub bill, cancellable at the end of any cycle, and an upgrade from Pro to Pro+ charges only the price difference for the rest of the month.

What changed, and what the old numbers meant

Cursor's previous ladder, Hobby, Pro at $20, Pro+ at $60 and Ultra at the same price as today, with Teams seats at $40 and $120, continues for existing customers; new customers start on the X Subscription plans above, per Cursor's legacy plans page. The old Pro included an allowance for third-party models and maps to Premium, so a Claude user who switches pays ten dollars more a month for the plan and loses the included Claude usage.

Copilot moved from premium requests to AI credits. The $5 and $31 figures you may see next to Pro and Pro+ on GitHub's page, and the one next to Max, are the flex allotments on top of the base credits, in dollars, not plan prices, and the $19 figure is the Business seat. The subscriptions are the Pro, Pro+ and Max prices in the table above.

For the models behind both tools, our Claude pricing, ChatGPT pricing and Grok pricing pages cover the consumer plans, and LLM API prices 2026 lists the per-token rates that both of these subscriptions pass through. If a chat window is enough for your coding, Claude vs ChatGPT compares them head to head, and if you are not a developer at all, Bolt vs Lovable covers the app builders that write the code for you.

Who should pick which

Pick GitHub Copilot if you want Claude, GPT or Gemini models and a predictable bill: Pro for light use, Pro+ once agent sessions become daily, Max when agents run most of the day. Pick it too if your team needs IP indemnity or SSO on a list price, or if you work in JetBrains, Visual Studio, Xcode or Neovim rather than VS Code.

Pick Cursor if you want its editor, its cloud agents and Bugbot in one place, and you are content to run Grok and Composer as your main models: Premium is then a flat price with a weekly reset that forgives a bad week. Pick Plus only for Grok-only work, since it has no cloud agents, no Bugbot and no third-party models. Skip Super and Ultra if your usage is mostly Claude or GPT, because those tiers do not reduce the third-party bill.

The adoption data points the same way. In JetBrains' Developer Ecosystem Survey 2026, run with more than 15,000 professional developers in May to July 2026, 21% used GitHub Copilot at work, down from 29% a year earlier, and 12% used Cursor, down from 18% in January; Claude Code led at 39%. Neither tool is the default choice any more, which is a reason to buy on price and model fit rather than on reputation.

How we compared

Louis Corneloup, founder of Dupple and Toolradar, wrote this comparison from the vendors' own pages, read on October 26, 2026: Cursor's pricing page, its models and pricing, team pricing and legacy plans documentation, GitHub Copilot's plans page, and GitHub's documentation on plans, usage-based billing and model pricing. The cost scenarios are our arithmetic on those published rates. The 179-tool count is from Toolradar's live AI coding ranking that month, and the adoption shares are JetBrains' own. More on how we publish is on the AIpresso about page.

FAQ

Which is cheaper in 2026, Cursor or GitHub Copilot?

GitHub Copilot is cheaper for anyone who wants Claude, GPT or Gemini models: Pro includes $15 of model credits, Pro+ includes $70, and Max doubles its price into credits. Cursor is cheaper only for Grok and Composer users, where Plus and Premium are flat prices; Claude and GPT cost list price on top of any Cursor plan.

How much does Cursor cost per month?

Cursor's plans are Free at $0, Plus at $8 a month ($80 a year), Premium at $30 ($300 a year), Super ($1,000 a year, with 15x the Plus pool) and Ultra at $200 ($2,000 a year), excluding tax. Teams seats cost $30, $100 or $200 per user. Third-party models such as Claude and GPT are not included and are billed at API list price from prepaid credit on Premium and above.

How much does GitHub Copilot cost per month?

GitHub Copilot costs $0 on Free, $10 on Pro, $39 on Pro+ and $100 on Max for individuals, and $19 per seat on Business or $39 per seat on Enterprise for organizations. Each paid plan includes GitHub AI Credits worth $0.01 each: 1,500 on Pro, 7,000 on Pro+ and 20,000 on Max, including a flex allotment GitHub may adjust.

Can I use Claude in Cursor and in GitHub Copilot?

Yes on both, at different prices. In Cursor, Claude models are available on Premium, Super and Ultra at API list price from a prepaid credit balance, and not at all on Free or Plus. In GitHub Copilot, Claude Sonnet 5.5, Opus 5.5 and Fable 5.1 are in the picker on every paid plan and draw from the included credits at the same per-token rates.

Do unused credits or usage roll over?

No on either. Copilot's included credits reset to the full monthly amount at 00:00 UTC on the first of each calendar month and unused credits are forfeited. Cursor's pool refreshes weekly, so unused usage from one week does not carry into the next, but a heavy day cannot wipe out a whole month.

Is there a free Cursor or GitHub Copilot plan?

Yes. Cursor Free gives limited Agent, Chat and Tab usage with no third-party models or cloud agents. Copilot Free gives 2,000 completions a month, a limited credit allowance, agent mode, the CLI and MCP, with the model chosen automatically. Students get Copilot free, and verified teachers and open-source maintainers get Copilot Pro free.

Cite this: AIpresso, "Cursor vs GitHub Copilot 2026: Pricing, Limits and Which to Pay For", November 2026.

AIpresso covers the AI tools worth your time in a 5-minute morning read. Join professionals who want the signal without the hype.

Some offers on this page may be paid placements or contain affiliate links.