Aanchal Parmar
Product Marketing Manager, Flexprice
Cursor is an AI code editor used heavily by developers for code generation and background agents. As of April 2026, its pricing has multiple components: monthly plans, usage-based model access, on-demand overages, and add-on products like Bugbot.
For developers, unclear pricing isn't a minor inconvenience, it affects whether you can trust the tool in production. Cursor's pricing has gone through multiple changes, most notably in June 2025 when it moved from request-based to usage-based billing. That shift triggered real concerns in its community and reshaped how the plans work today.
This breakdown explains Cursor's current pricing, why it's structured this way, and how to pick a plan based on actual usage, not just the pricing table.
How Cursor's pricing changed from request-based limits to usage-based billing
Cursor originally used request-based limits. The Pro plan included a set number of fast premium requests per month, unlimited slow premium requests, and unlimited Tab completions. Teams had per-user request caps with SSO, privacy mode, and admin features. Limits were based on the number of requests, not the compute cost of those requests.
In June 2025, Cursor replaced request caps with usage-based billing pegged to model API pricing. The change aligned billing with actual compute cost, heavier models, longer contexts, and MAX mode consume more of your included amount, while lighter use stays within it. The rollout, however, was rushed and poorly communicated, leading to backlash from existing users.
Since then, Cursor has continued upgrading its pricing. The current structure includes six tiers for its core editor (Hobby, Pro, Pro+, Ultra, Teams, and Enterprise) plus a separate Bugbot product for AI code reviews.
The backlash Cursor faced after the pricing change
When Cursor moved from request-based limits to usage-based billing in June 2025, the announcement was poorly timed and poorly explained. Existing customers reported being overcharged without prior warning, as the new system measured consumption in dollar-equivalent credits tied to model API costs, something many hadn't tracked before.
The main pain points:
“Unlimited" confusion, many assumed it applied to all models, not just Auto.
No clear usage meter, developers couldn't see in real time how much of their included amount was left.
Short notice, no extended transition window before charges kicked in.
By July 4, Cursor posted an apology, clarified the new rules, and offered refunds for unexpected usage between June 16 and July 4. The damage was partly reputational: on Reddit and Cursor's forum, threads remain where users share billing screenshots and warn others to watch their usage closely.
Check out this Reddit thread by Cursor
How Cursor's current pricing is structured and calculated
Cursor's current pricing uses a tiered model across individual and business plans. Every plan includes a set amount of model usage. On-demand usage allows you to continue using models after your included amount is consumed, billed in arrears.
Individual Plans:
Hobby (Free): No credit card required. Limited Agent requests and limited Tab completions.
Pro ($20/mo): Everything in Hobby, plus extended limits on Agent, access to frontier models, MCPs, skills, and hooks, and cloud agents.
Pro+ ($60/mo, Recommended): Everything in Pro, plus 3x usage on all OpenAI, Claude, Gemini models.
Ultra ($200/mo): Everything in Pro, plus 20x usage on all OpenAI, Claude, Gemini models and priority access to new features.
Business Plans:
Teams ($40/user/mo): Everything in Pro, plus shared chats, commands, and rules, centralized team billing, usage analytics and reporting, org-wide privacy mode controls, role-based access control, and SAML/OIDC SSO.
Enterprise (Custom): Everything in Teams, plus pooled usage, invoice/PO billing, SCIM seat management, AI code tracking API and audit logs, granular admin and model controls, and priority support and account management.
How usage works:
Every plan includes a set amount of model usage.
On-demand usage allows you to continue using models after your included amount is consumed, billed in arrears.
Pro+ gives you 3x and Ultra gives you 20x the usage on all OpenAI, Claude, Gemini models compared to Pro.
When privacy mode is enabled, Cursor guarantees that code data is never stored by their model providers or used for training.
Self-serve plans support all major credit and debit cards. For invoice-based billing and wire transfers, you need to contact Cursor for an Enterprise plan.
Get started with your billing today.
How on-demand usage works once you exceed Cursor's included amount
Every paid plan comes with a set amount of model usage included in the monthly fee.
Pro: Base included amount.
Pro+: 3x usage on all OpenAI, Claude, Gemini models.
Ultra: 20x usage on all OpenAI, Claude, Gemini models.
When you exceed your included amount, on-demand usage kicks in. This allows you to continue using models, billed in arrears. Heavier models and longer contexts cost more, so your included amount runs out faster if you use those often.
If you finish your included amount before the month ends, you have two choices:
Switch to Auto, which is included with Pro/Pro+/Ultra, but may not always run on the largest model.
Continue with on-demand usage, billed in arrears at the model's cost.
Example: If you spend most of your time in MAX mode, a single request might use several times more of your included amount than a normal request. If you mostly use smaller models, you'll stretch your included amount much further.
How models and features affect what you can use and what you pay
Not all Cursor features work the same way, or cost the same to use.
Auto vs frontier models:
Auto is included with Pro, Pro+, and Ultra. Cursor chooses the model for you, balancing speed and cost.
Frontier models (accessible on Pro and above) use your included amount and trigger on-demand billing if you go over.
MAX mode:
Gives you a bigger context and longer reasoning, but uses more of your included amount per request.
The bigger the context, the faster your included amount runs out.
BYO API key limits:
You can connect your own API key for some models to avoid using your included amount.
But key features, like Tab and Apply from Chat, run on Cursor's own models and can't be billed to your key.
MCPs, skills, and hooks (Pro and above):
Pro and above plans include support for MCPs, skills, and hooks, extending Cursor's capabilities with external tools and custom workflows.
Cloud agents (Pro and above):
Pro and above plans include cloud agents, letting you run AI tasks in the cloud without tying up your local machine.
Why this matters:
If most of your work is Tab or Auto, you can stay within Pro's limits. If you use frontier models heavily or run MAX mode often, expect to need Pro+ or Ultra for the extra headroom.
How to choose the right Cursor plan for your workflow
Picking the right plan comes down to how often you use frontier models and how heavy your requests are.
If you mostly use Tab and Auto:
Go with Pro ($20), you'll have extended limits on Agent and access to frontier models, which cover most standard development workflows.
If you use frontier models regularly but don't need the full Ultra headroom:
Go with Pro+ ($60), you get 3x usage on all OpenAI, Claude, and Gemini models. This is Cursor's recommended tier for a reason: it's the sweet spot for active developers.
If you run MAX mode or frontier models all day:
Choose Ultra ($200), 20x usage on all OpenAI, Claude, Gemini models covers heavy, high-context work, plus you get priority access to new features.
If you're part of a team needing shared workflows and admin controls:
Use Teams ($40/user), everything in Pro plus shared chats, commands, and rules, centralized team billing, usage analytics and reporting, org-wide privacy mode controls, role-based access control, and SAML/OIDC SSO.
If you need enterprise-grade controls and billing:
Go with Enterprise (custom pricing), everything in Teams plus pooled usage, invoice/PO billing, SCIM seat management, AI code tracking API and audit logs, granular admin and model controls, and priority support and account management.
If you want predictable cost above all else:
Stick to Auto and avoid on-demand overages, or use BYO API key where possible.
The easiest way to decide: Track your model usage for a week. If you're burning through the Pro's included amount in days, Pro+ or Ultra will prevent constant on-demand charges.
Hidden costs and upgrade traps to watch out for
Cursor's pricing table doesn't always show the extra ways you can spend more than you expect.
BYO API key limitations
You can connect your own API key for some models to bypass your included amount.
But key features like tab and apply from chat always run on Cursor's own models, so they can't be billed to your key.
Background Agents
Billed separately at the model's API rate.
No fixed monthly cap, you need to set your own spend limit.
Bugbot subscription
Not included in any Cursor editor plan. It is a separate product with its own pricing.
Bugbot Pro costs $40/user/mo (14-day individual trial, reviews on up to 200 PRs/mo, access to Bugbot rules).
Bugbot Teams costs $40/user/mo (14-day team trial, code reviews on all PRs, analytics and reporting dashboard, advanced rules and settings).
Bugbot Enterprise has custom pricing (30-day org-wide trial, advanced analytics and reporting, priority support and account management).
MAX mode
Larger context windows and deeper reasoning use more of your included amount per request.
Heavy MAX usage can burn through your included amount in days, even on Ultra.
On-demand usage
Once your included amount is consumed, on-demand usage kicks in automatically, billed in arrears.
Without tracking, it's hard to judge when you're nearing that threshold.
Upgrade pressure
If you regularly hit your included amount early, you may feel pushed toward Pro+, Ultra, or Teams.
Without real-time usage tracking, it's hard to judge when you're nearing that point.
Knowing these helps you avoid unexpected charges and choose a pricing plan based on actual, not assumed, usage.
How to clone Cursor’s pricing with Flexprice’s Prompt to Plan Feature
If Cursor's pricing model resonates with how you want to charge for your own product, you don't have to build the billing infrastructure from scratch. Flexprice's MCP server lets you replicate this exact structure, fixed plans, usage credits, and pay-as-you-go overages, without writing billing logic yourself. Let's see how this works:
Step 1:
First, visit our website Flexprice.io, then click on Try For Free button to sign up
After that, a new window will open where you can fill in your credentials to get started with us
Step 2:
After you're signed up, the Flexprice dashboard will open up. Now, navigate to the product catalog, there you can see plans. Click on that
Step 3:
Click on Product Catalog from the left side of the nav bar and then click on Plans.
Step 04:
On the top right corner you can see a button CREATE WITH AI, so click on that.
Step 05:
Once you click on that you can see there are many templates; you just need to select the Cursor, and then the prompt for Cursor pricing will appear in the chat. Then click enter and the analyzing will begin.
Step 06:
After analyzing is over a new window with all the features and paln will come,now click on create
Step 07:
After this process is completed you can see all the plans and features have been created according to Cursor pricing.
How to clone Cursor’s pricing with Flexprice’s MCP server
Step 01:
Ensure that you are successfully set up with Flexprice's MCP server. You can check it by going into settings, then into developer options. There, you can see the Flexprice MCP server status
Step 02:
Access Claude or any other AI tool that integrates with the MCP server.Then provide a single prompt asking the AI tool to analyze Cursor’s pricing page and replicate its entire structure through the Flexprice MCP. In the prompt, include Cursor’s pricing link.
Step 03:
Navigate to your Flexprice dashboard to review the newly created assets. You will find that the AI has used the context from Cursor’s pricing page to automatically pre-fill and create plans.
And that's all it takes to replicate Cursor's pricing structure using Flexprice’s MCP server. Both methods get you to the same result, the only difference is how hands-on you want to be.
The AI template is faster if you want a one-click starting point, while the other approach gives you more control through your own AI assistant.
Once everything is set up in your dashboard, you can adjust the plan names, credit limits, and overage rates to better fit your product. The hard part is already done, now your pricing just needs to learn your product's name.
Balancing flexibility and predictability in Cursor's pricing
Cursor's tiered model offers flexibility; you can use any supported model, scale context size, and pick a tier that matches your usage intensity. With six plan options across individual and business tiers, plus a separate Bugbot product, there's a configuration for most workflows.
The trade-off is predictability. If you don't track which models you use or how often you're in MAX mode, on-demand costs can jump unexpectedly. The June 2025 rollout showed that even well-intentioned changes can frustrate users if the rules aren't clear and the tools to monitor usage aren't in place.
For most developers, the best approach is to start small, track your usage for a month, and only move up when you know exactly why you need the extra headroom. Flexibility is valuable, but only if you can see, predict, and justify the cost.
How does Cursor's usage-based pricing work, and how is it different from request-based billing?
What is Cursor's on-demand pricing, and how can I avoid unexpected overage charges?
Which Cursor plan is best for developers who use AI coding tools heavily every day?
How does Cursor's pricing compare to other pay-as-you-go AI coding tools?
Can I replicate Cursor's usage-based pricing model for my own SaaS or AI product?
