If you’re a developer, you’ve probably heard the buzz around Anysphere’s Cursor. It’s an AI-powered code editor that’s been making waves. But you might have also heard about the recent confusion and frustration from users over changes to Anysphere pricing.
Let’s be real: trying a new tool is one thing, but getting hit with a surprise bill you weren’t expecting is something else entirely. In this guide, we’ll break down Cursor’s plans without the jargon, get into what happened with their recent pricing controversy, and talk about what it all means for businesses looking to invest in AI tools.
What are Anysphere and Cursor?
First off, let’s clear up the names. Anysphere is the research lab, and Cursor is their main product. Basically, Cursor is a code editor designed from the ground up to work with AI. It helps developers write, debug, and refactor software a whole lot faster.
It integrates directly with big language models from places like OpenAI, Anthropic, and Google. This means you get AI help, from spitting out boilerplate code to fixing tricky bugs, right inside your editor, so you don’t have to constantly switch between windows.
People loved it. Cursor saw a huge surge in users, from solo developers to big companies, reportedly jumping to over $500 million in annual recurring revenue. But that rapid growth brought some problems, especially when it came to their pricing.
A complete breakdown of Anysphere pricing for Cursor in 2025
Cursor has a few different plans for different kinds of users, from students tinkering on the side to large enterprise teams. Let’s walk through each one, based on their official pricing page.
Anysphere pricing for individual and business plans
Here’s what you can choose from:
-
Hobby Plan (Free): This is where you start. It’s free and gives you a limited number of AI requests, so it’s good for casual developers or anyone who just wants to see what Cursor is all about. You also get a two-week trial of the Pro plan to get a feel for the paid features.
-
Pro Plan ($20/month): This is their most popular plan and the one that caused all the recent trouble. It’s for individual professional developers and gives you access to all the AI models. We’ll get into the details of the recent changes to this plan in just a bit.
-
Pro+ Plan ($60/month): A step up from Pro, this plan gives you three times the usage on all the AI models. It’s for developers who find they’re consistently hitting the limits of the standard Pro plan.
-
Ultra Plan ($200/month): This is the top tier for "power users" who need predictable, high-volume access to AI. It comes with 20 times the usage of the Pro plan and you get first dibs on new features.
-
Teams Plan ($40/user/month): This plan is built for businesses. It includes everything in the Pro plan but adds key features for working in a team, like one bill for everyone, admin controls, and single sign-on (SSO).
-
Enterprise Plan (Custom): For big companies with specific needs, this plan offers custom pricing. It comes with features like pooled usage across the company, invoice billing, SCIM for managing users, and a dedicated support contact.
Here’s a simple table to see how the plans compare:
| Feature | Hobby | Pro | Pro+ | Ultra | Teams | Enterprise |
|---|---|---|---|---|---|---|
| Price | Free | $20/mo | $60/mo | $200/mo | $40/user/mo | Custom |
| Target User | Individuals / Students | Professional Developers | Power Users | Heavy Power Users | Small to Mid-Sized Teams | Large Organizations |
| AI Model Usage | Limited requests | Usage-based credits (~$20 value) | 3x Pro usage | 20x Pro usage | Same as Pro, per user | Pooled usage |
| Key Features | Basic AI chat, Two-week Pro trial | Access to all models, Tab completions | Everything in Pro | Everything in Pro, priority access | Centralized billing, Admin controls, SSO | Everything in Teams, Invoice billing, SCIM, Priority support |
| Billing Model | Free | Monthly | Monthly | Monthly | Per User, Monthly | Custom / Annual |
The Pro plan controversy: Why the Anysphere pricing change was so confusing
The user backlash wasn’t just about a simple price increase. The real problem was a complete change in how the popular $20/month Pro plan was billed, a change that many users felt was communicated poorly.
The pricing change: From a fixed request model to usage-based credits
Originally, the Pro plan was predictable. For $20 a month, you got 500 "fast" requests on the good models. If you burned through those, you could still make unlimited "slow" requests. You knew exactly what you were getting for your money.
Then, things changed. The plan switched to giving users $20 worth of usage credit per month, charged at the direct API rates of the AI models. All of a sudden, your $20 wasn’t buying you a set number of requests anymore; it was buying a variable amount of processing power.
As covered by outlets like TechCrunch and Dataconomy, users were completely caught off guard. They found their credits vanishing way faster than they expected, especially when using powerful (and expensive) models like Anthropic’s Claude Opus 4.
The impact on users: Unexpected costs and limits
For developers who relied on Cursor every day, the effects of this change were immediate and frustrating.
-
Rapid credit depletion: A developer could easily burn through their entire month’s $20 credit with just a few complex queries. This left them locked out of the tool for the rest of the month unless they were willing to pay more.
-
Surprise bills: The real shock came for users who hadn’t set a hard spending limit. They were automatically billed for any usage over the initial $20. Some people reported getting bills for over $70 for just a single day’s work.
-
Loss of predictability: The flat-rate fee that used to guarantee a full month of access was gone. This made it nearly impossible for freelance developers and businesses to budget for the tool.
To be fair, Anysphere’s CEO, Michael Truell, did issue an apology for the confusing rollout and offered refunds. But the whole incident was a wake-up call for many about the tricky nature of AI pricing.
This video discusses the controversy around Cursor's abrupt pricing change and the subsequent user backlash.
What volatile Anysphere pricing means for your business
The Cursor situation isn’t just some one-off event; it’s a perfect example of a bigger headache facing any business that uses AI tools. It highlights just how important it is to pick the right vendors and really understand what you’re paying for.
How underlying AI models affect pricing
A lot of the AI tools out there today, including Cursor, are built using foundational models from a handful of big players like OpenAI, Anthropic, and Google. As these models get smarter, they also get more expensive to run.
For instance, Fintech Weekly pointed out that Anthropic’s Claude Opus 4 can cost $15 for every million input tokens (roughly 750,000 words) and a whopping $75 for every million output tokens. Those costs get passed on from companies like Anysphere directly to their customers, as we just saw.
The challenge of unpredictable pricing and usage-based billing
Usage-based pricing might sound fair, "you only pay for what you use", but it creates a lot of risk and operational stress for a business.
Trying to forecast your monthly software budget becomes a total guessing game. How can you plan your expenses when your bill changes depending on how many support tickets your team gets or how many complex code questions your developers ask? This model forces teams to constantly watch their usage just to avoid getting their service cut off or getting a massive bill at the end of the month. It’s a distraction from the work that actually matters.
Beyond Anysphere pricing: Why predictable AI pricing matters for support teams
This is where a different approach to AI pricing is really needed, especially for crucial parts of a business like customer support, where you absolutely need a predictable budget.
eesel AI’s transparent, value-based model
eesel AI is an AI platform for customer service that was designed specifically to avoid this problem. Instead of a confusing usage-based system, eesel AI has a straightforward and predictable pricing model.
Here’s the main difference: there are no per-resolution or usage-based fees. Our plans are based on a fixed number of "AI interactions" per month (an interaction is just one AI reply or one AI action). That means you will never get a surprise bill, no matter how busy your support team is. All of our core products, including our AI Agent, AI Copilot, and AI Triage, are included in every plan, so you get a ton of value without any hidden fees.
A screenshot showing eesel AI's transparent, value-based pricing plans, which contrasts with the unpredictable Anysphere pricing model.Go live in minutes and test with confidence
The other big difference is how you get started. While Cursor users were surprised by costs after they started using the tool, eesel AI lets you test everything out and remove the risk before you commit.
With eesel AI’s powerful simulation mode safeguards your budget, you can test your AI setup on thousands of your own past support tickets. This gives you a clear picture of how the AI will perform and how much it will save you before you spend a single dollar. It takes the guesswork out of the equation and makes sure you’re on the right plan from the start. You can even set it all up yourself in a few minutes, no mandatory sales call required.
The eesel AI simulation mode allows users to test the AI's performance on past tickets, providing clear ROI before committing, a safer approach than the surprising Anysphere pricing changes.Choosing the right AI pricing model for your team
The recent drama around Anysphere pricing is a good lesson for any team looking to adopt AI. While usage-based tools can be extremely powerful, their complicated and unpredictable pricing can create financial uncertainty that most businesses just can’t afford.
For the parts of your business that you rely on every day, a clear and predictable pricing model isn’t just a nice-to-have. It’s what allows you to operate smoothly, scale with confidence, and sleep better at night.
Looking for an AI partner you can count on?
The world of AI is complicated enough without having to worry about surprise bills. eesel AI offers powerful, easy-to-use AI for customer support with simple, predictable pricing.
Run a risk-free simulation on your own data and see exactly how much time and money you can save. Try eesel AI for free.
Frequently asked questions
Cursor offers several plans, including a free Hobby plan, individual Pro, Pro+, and Ultra plans, and business-focused Teams and custom Enterprise plans. Each plan caters to different user needs, from casual developers to large organizations.
The controversy arose when the $20/month Pro plan switched from a fixed number of "fast" requests to a usage-based credit system. Users were unexpectedly charged more as their $20 credit was quickly depleted by direct API costs of powerful AI models, leading to surprise bills and a loss of predictability.
Usage-based Anysphere pricing makes it challenging for businesses to forecast monthly software expenses accurately. The variable cost, dependent on the volume and complexity of AI interactions, can lead to unpredictable bills and requires constant monitoring, diverting focus from core tasks.
Cursor’s Anysphere pricing is heavily influenced by the cost of the foundational AI models it integrates, such as those from OpenAI and Anthropic. More advanced models are more expensive to run, and these costs are passed directly to users, particularly in the usage-based billing structure.
Yes, Anysphere offers a "Teams Plan" at $40/user/month for small to mid-sized businesses, which includes centralized billing and admin controls. For larger enterprises, there’s a custom "Enterprise Plan" with pooled usage, invoice billing, and dedicated support.
To avoid surprise bills with Anysphere pricing, users on usage-based plans should carefully monitor their AI credit consumption and set hard spending limits within their account settings. Understanding the per-token costs of different AI models can also help manage expectations.
