Pricing guide
Cursor Pricing (2026): Hobby, Pro, Pro+, Ultra and Teams Plans and Usage Costs
Every current Cursor plan, what it costs, and how usage-based billing changes your real bill. All figures read from Cursor's own pricing page in July 2026, with the flat-price way to run an agent on your repo alongside.
No card to start · Review-first: the agent never merges
Pick a task
Plan
- planning
Files changed
Test run
Pull request
You review and merge. Agentcode never merges on its own.
The short answer
Cursor pricing in July 2026 runs from $0 to $200 a month for individuals: Hobby ($0, limited), Pro ($20 a month, about $16 annual), Pro+ ($60 a month), Ultra ($200 a month), and Teams at $40 per user. Since June 2025 every plan is usage-based: each one includes a set amount of model usage, and heavy Agent or MAX-mode work past that allowance bills as on-demand overage in arrears, pegged to model API cost. Tab autocomplete rarely moves the meter, but agent runs do, so a busy month costs more than the sticker price. The flat-price way to run an agent on your repo, with no usage meter, is Agentcode at $29 a month.
Cursor plans and prices
Six plans, individual and organization. The column that decides your real cost is the included usage: that is the allowance Agent and MAX-mode runs draw from before you start paying on-demand overage.
| Plan | Price | Included usage | Best for |
|---|---|---|---|
| Hobby | $0 free | Small included pool | Trying the editor out. |
| Pro | $20 per month | Included model usage | An individual developer. |
| Pro+ | $60 per month | ~3x Pro usage | A developer who lives in the Agent. |
| Ultra | $200 per month | ~20x Pro usage | The heaviest individual users. |
| Teams | $40 per user / mo | Pooled, plus $120 Premium seat | A team that needs admin and SSO. |
| Enterprise | Custom contact sales | Pooled usage | A larger org with procurement needs. |
Prices and usage allowances as published by Cursor and checked in July 2026. Annual billing takes about 20 percent off each paid plan. Cursor changes this pricing often, so confirm on cursor.com before you buy.
What each Cursor plan gets you
Hobby
$0 · free · Small included poolNo card required. You get a limited number of Agent requests and limited Tab completions a month, then it stops until the next cycle. Enough to feel the editor and the Tab model, not enough for daily professional work.
Pro
$20 · per month · Included model usageExtended Agent limits, generous Tab completions, access to frontier models, MCP, skills and hooks, and cloud agents. Billed at about $16 a month on the annual plan (20 percent off). The included model usage is where a heavy month starts to run past the sticker price.
Pro+
$60 · per month · ~3x Pro usageEverything in Pro with roughly three times the included usage across OpenAI, Claude and Gemini models. Aimed at people whose day is mostly agent runs and long-context work rather than autocomplete.
Ultra
$200 · per month · ~20x Pro usageAbout twenty times the included usage of Pro plus priority access to new features and models. This is the plan for someone running the Agent almost continuously who would otherwise pay large overage on Pro.
Teams
$40 · per user / mo · Pooled, plus $120 Premium seatEverything in Pro plus centralized billing, shared rules and skills, usage analytics, org-wide privacy mode and SAML/OIDC SSO. A $120 Premium seat added in June 2026 carries about five times a Standard seat's included usage for the heaviest team members.
Enterprise
Custom · contact sales · Pooled usageEverything in Teams plus pooled usage across seats, invoice and PO billing, SCIM seat management, access controls for repositories, models and MCP, audit logs and priority support.
How usage-based billing changes the bill
The plan price is only half the number. Since June 2025 Cursor meters the parts most developers actually adopt it for, and understanding that meter is the difference between the price you expect and the price you pay.
Included usage, then overage in arrears
Cursor dropped fixed request caps in June 2025 for usage-based billing pegged to model API cost. Each plan bundles a set amount of model usage; once you exhaust it, on-demand usage keeps working and bills in arrears at model rates, up to a spend limit you set. Heavier models, longer contexts and MAX mode burn the allowance faster than light editing does.
Tab is cheap, the Agent is not
Cursor's Tab autocomplete and next-edit suggestions rarely move the meter, which is why a developer who mostly writes their own code and accepts completions can stay comfortably inside Pro. What draws usage down fast is the Agent: multi-file edits, long-context reasoning and MAX mode. Someone running the Agent across a codebase all day can outspend the included pool well before month end.
Model a busy month, not an average one
Because Agent runs drive the cost, the honest way to budget Cursor is to price a heavy sprint rather than a typical week. Estimate how many agent tasks a productive developer triggers near a deadline, add the overage past the included allowance, and that is the figure a finance team should approve. It is often higher than the plan price implies, which is why the Pro+, Ultra and Premium seats exist.
Cursor vs a flat-priced PR agent
Cursor is an AI-first editor with cloud agents bolted on; Agentcode is an async agent that opens pull requests and bills flat. They overlap less than the category name suggests, so here is where each one is genuinely ahead.
| Cursor | Agentcode | |
|---|---|---|
| Billing model | Subscription plus usage-based billing pegged to model cost, in arrears (since June 2025) | Flat subscription, no usage meter or per-task overage |
| Entry price | $20 / mo (Pro), about $16 annual | From $29 / mo, flat |
| What a busy month costs | More: Agent and MAX-mode runs draw down the included pool, then bill overage | The same as a quiet month |
| In-editor autocomplete (Tab) | Yes, its Tab model is best in class | No, this is an async PR agent, not an autocomplete |
| Model choice | Yes, pick OpenAI, Claude, Gemini and more per request | Managed model selection, no model to configure |
| Opens pull requests | Cloud agents open PRs, with model API billed on top | Yes, reviewable PRs on GitHub and GitLab, no metered top-up |
| GitLab support | Edits any local repo; cloud agents are GitHub-centric | Native GitHub and GitLab |
| Runs your test suite before the PR | In agent mode, within its sandbox | Yes, runs your existing suite and reports results in the PR |
| Trains on your code | No when Privacy Mode is enabled | Never, on any plan |
Cursor wins clearly on in-editor experience: its Tab model and per-request model choice are the best in the category. Where a flat plan wins is predictability and reach: Agentcode runs the same async PR workflow on GitHub and GitLab for one monthly price, so a busy month and a quiet one cost the same. Many teams run both, autocomplete from one and async PRs from the other. For the full breakdown see the Cursor alternative page and the flat vs metered pricing comparison across every tool.
Cursor pricing: your questions answered
How much does Cursor cost?
Cursor costs $0 on the Hobby plan, $20 a month for Pro (about $16 on annual billing), $60 a month for Pro+, $200 a month for Ultra, and $40 per user a month for Teams, as of July 2026. Every paid plan includes a set amount of model usage, and heavy Agent or MAX-mode use past that allowance bills as on-demand overage on top of the subscription.
Is Cursor free?
Cursor has a free Hobby tier, but it is limited. You get a small monthly pool of Agent requests and Tab completions, then it pauses until the next cycle. It is enough to try the editor and its autocomplete, not enough for daily professional work, which is what the $20 Pro plan and the $40-per-seat Teams plan are for.
Why did Cursor change its pricing?
In June 2025 Cursor replaced fixed request caps with usage-based billing pegged to model API cost. The goal was to align what you pay with the compute you actually use: heavier models, longer contexts and MAX mode draw down your included usage faster, while lighter work stays inside it. The tradeoff is that your bill now moves with how hard you run the Agent.
What is the difference between Cursor Pro and Ultra?
Cursor Pro is $20 a month with a standard included-usage allowance. Ultra is $200 a month with roughly twenty times that allowance plus priority access to new features and models. Pro+ sits between them at $60 for about three times Pro usage. The right tier depends on how many Agent runs you trigger: Ultra only pays off if you would otherwise rack up heavy overage on Pro.
Does Cursor charge per use?
Partly. The subscription is flat, but on top of it Cursor meters model usage: each plan includes an allowance, and once you exhaust it, on-demand usage bills in arrears at model API rates. Tab completions and light editing rarely move the meter; Agent runs, MAX mode and long-context requests are what draw it down fastest, so a productive sprint can cost more than a quiet week.
Is Cursor worth it?
For in-editor AI, Cursor is one of the strongest tools available: its Tab autocomplete and per-request model choice are excellent, and $20 Pro is a fair entry price. The number to model before you commit is the usage meter, because a heavy Agent month can push the real cost well past $20. If you want a predictable flat bill and an agent that opens pull requests rather than an editor, it is worth comparing a flat-priced PR agent.
Is there a flat-price alternative to Cursor?
Yes. Agentcode charges a flat fee from $29 a month with no usage meter and no per-run overage: you describe a task, it plans, edits the code, runs your test suite and opens a reviewable pull request on GitHub or GitLab, for the same price whether you run it once or fifty times. It is an async PR agent rather than an editor, so it complements Cursor for teams that want in-editor autocomplete and predictable async runs both.
Last updated: July 2026
Weighing Cursor against the field?
These pages compare pricing and reach tool by tool: Cursor alternative, GitHub Copilot pricing, Claude Code alternative and the Agentcode vs Cursor comparison.
Or step back to AI coding tool pricing across every tool, the best AI for coding in 2026, or what an AI coding agent does before you weigh the bill.
A flat bill, on your repo
Connect a GitHub or GitLab repository, describe a task, and read the pull request it opens. Same price whether you run it once or a hundred times, with no usage meter to watch.