AI is everywhere these days, and one platform shift that’s happening at an eye-watering pace is the adoption of AI-enabled IDEs. If you’re reading this article, you’re probably a user of things like Cursor, Claude Code, or heck….maybe you’re one of the users I’ve yet to meet using Windsurf.

The growth of these platforms is incredible. Taking Cursor for example: the company incorporated in 2022 and supposedly is nearing an annualized run-rate of $1B dollars coming from millions of users around the world.

While the revenue is impressive to Cursor, it's ultimately a cost for you. This article is meant to be an explainer of how Cursor charges, unpack some of the drama of the company’s recent pricing changes, and present a path forward for how to account for these costs at your organization moving forward.

How Does Cursor Pricing Work?

Cursor has two classes of plans: Individual and Team plans. Both classes have a similar pricing structure: a monthly subscription fee and a usage-based fee for requests and corresponding token usage whenever you make a request.

Public pricing subscriptions are listed below and are paid monthly. You can also commit at an annual rate upfront if you’re so inclined to receive a subscription discount of 20%. Each plan has a set of request limits that are advertised as being “fast requests”. Once you exhaust your request limit, you have two options:

• Upgrade your plan to pay-as-you-go usage-based pricing -or-
• Have degraded service to “slow requests”. So even if you’re capped out on your request limit for the month, you’ll be dropped into a pool to serve slower requests. This is essentially so that Cursor has some protections on its systems for customers who don’t upgrade.

Individual Subscription Rates

Plan Name	Price per month	Limits
Hobby	Free	50 requests
Pro	$20	500 fast requests
Ultra	$200	500 fast requests

Cursor individual subscription rates

Team Subscription Rates

Plan Name	Price per month	Notes
Team	$40	Base seat cost, includes usage pools shared across the team
Enterprise	Inquire	Negotiated pricing with dedicated support + custom SLAs

Cursor team subscription rates

Consumption-Based Token Usage

Beyond the subscription tiers, Cursor charges for token usage. This is where costs can start to climb, especially if you or your team are using Cursor as a daily driver. Every request is ultimately backed by API calls to foundation models (like OpenAI’s GPT-4 or Anthropic’s Claude), and Cursor passes some of that cost downstream.

• Input tokens: The size of your prompt (code context, docstrings, etc.).
• Output tokens: The length of the generated response.
• Effective pricing: Varies by plan, but generally you’ll see effective rates of ~$0.30–$0.60 per million tokens depending on which underlying models you’re hitting.

Cursor abstracts most of this away with its “fast request” buckets, but once you exceed those, you’re effectively paying for raw token consumption, and at scale, this can add up quickly.

The Recent (2025) Pricing Controversy

Cursor recently announced changes that reduced the “fast request” allotments for Pro and Ultra users while nudging customers toward pay-as-you-go overages. For individual developers, this felt like a rug-pull - the Pro plan, which used to feel unlimited for many, suddenly started rate-limiting heavy users.

From Cursor’s perspective, this makes sense. As more users rely on higher-cost foundation models, their margins compress unless they align pricing closer to real usage. Still, the abrupt change created a fair amount of friction in the developer community and sparked conversations about whether dev tools should monetize like infrastructure platforms (AWS, GCP or like productivity apps (Notion, Figma).

What This Means for You

If you’re an individual developer, you can likely stick with Pro until you notice friction. For most hobbyists and light professional use, the $20/month tier provides a lot of value. But if Cursor has become central to your workflow (daily pair programming, frequent large context requests), expect to budget for overages or Ultra.

For organizations, the story is different:

• Forecast usage: Look at how many requests per engineer per day you expect. A 10-person engineering team easily burns through thousands of requests monthly.

• Track token-heavy workflows: Large codebases, refactor requests, and natural-language Q&A against your repos are all expensive. Soon, Vantage will be able to ingest Cursor costs to help out with this as well.

• Negotiate early: If Cursor is becoming mission-critical, push for Enterprise pricing to avoid unexpected overages.

Wrapping Up

Cursor is an incredible tool, but like many AI-enabled platforms, its pricing is still evolving as usage patterns mature and underlying model costs shift. The key takeaway is this: what feels like “just a dev tool” is in fact infrastructure spend hiding in plain sight.

As with cloud costs, understanding the levers (subscription tiers, request limits, token overages) and monitoring usage patterns will put you in a much better position to budget, and avoid unpleasant surprises.

At Vantage, we’ve seen this story before. Whether it’s AWS, GCP, Datadog, or now AI-native tools like Cursor, costs tend to start small and then scale dramatically as adoption grows. Treat Cursor like any other line item of cloud infrastructure: monitor it, forecast it, and be proactive in how you negotiate.