Cursor

Cursor pricing combines a flat monthly subscription with token usage billed at each model's API rate.

Pricing Model:

Pricing Model:

Hybrid: subscription plus metered usage

Hybrid: subscription plus metered usage

Hybrid: subscription plus metered usage

Packaging Model:

Packaging Model:

Freemium, Good / Better / Best (GBB)

Freemium, Good / Better / Best (GBB)

Freemium, Good / Better / Best (GBB)

Credit Model:

Credit Model:

Two pools, reset each billing cycle, no rollover documented

Two pools, reset each billing cycle, no rollover documented

Two pools, reset each billing cycle, no rollover documented

Updated on:

Cursor pricing: what you actually pay for

Cursor pricing combines a flat monthly subscription with token usage billed at each model's API rate. Every paid plan carries two separate usage pools that reset with your billing cycle, and once you spend them Cursor keeps working and bills the rest as on-demand usage. The headline number is easy. The part that decides your bill is which model you pick, because model choice sets how fast the pools drain.

Key takeaways

  • Individual plans run $20, $60 and $200 per month, and all three include both usage pools.

  • Usage is metered in tokens at each model's published API rate, not in an abstract credit.

  • Teams and Enterprise pay a Cursor Token Rate of $0.25 per million tokens on top of third-party model rates, including on bring-your-own-key usage.

  • Cursor publishes its own spend guidance: daily agent users typically reach $60 to $100 a month.

Cursor pricing in 2026

Plan

Price

Cursor Models pool

Other Models pool

 

Start (India only)

₹649/mo, tax inclusive

Included

Not included

Pro

$20/mo

Included

Included

Pro Plus

$60/mo

Included

Included

Ultra

$200/mo

Included

Included

Teams Standard

$40/user/mo

Included

Included

Teams Premium

$120/user/mo

Included

5x Standard agent limits

Enterprise

Custom

Included

Pooled usage

Source: cursor.com/docs/models-and-pricing, read 22 September 2026.

What Cursor actually meters

Cursor meters tokens, and it publishes the rate for every model it serves. This is worth stating plainly because Cursor's marketing pricing page does not say so. That page describes tiers as "extended Agent limits" and "3x Pro Agent limits" and never names a unit. The docs are a different story and disclose the whole rate card.

Usage splits across two pools that each reset with your monthly billing cycle:

  • Cursor Models covers first-party models: Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5. Composer 2.5 runs $0.50 per million input tokens and $2.50 output. Grok 4.6 runs $2 input and $6 output.

  • Other Models covers third-party models charged at the provider's API rate. Claude Sonnet 5 is $2 input and $10 output per million tokens. Claude Fable 5.1 is $10 input and $50 output.

Fast mode and long context multiply those rates. Composer 2.5 Fast bills $3 input and $15 output, six times the standard rate, and Grok 4.7 doubles past 256K input tokens, tripling if you combine the two. Start runs non-fast only, at a fixed medium effort level.

Rates differ by an order of magnitude across that list, so model selection is the single biggest lever on a Cursor bill.

Two surcharges sit on top. Teams and Enterprise accounts pay a Cursor Token Rate of $0.25 per million tokens on third-party models, and it applies to included usage, on-demand usage and bring-your-own-key usage alike. First-party Cursor models are exempt. On legacy request-based plans, Max Mode bills at the model's API rate plus 20%.

How the usage pools work

Both pools reset with your monthly billing cycle rather than accumulating. Cursor documents no rollover of unused pool balance and no expiry schedule beyond the cycle reset, so unspent included usage does not appear to carry forward.

The two pools stay separate. Spending your Other Models allowance does not touch the Cursor Models allowance, which is why Cursor positions its first-party models as the cheaper daily driver. The Start plan in India covers only the Cursor Models pool and has no access to the Other Models pool or to on-demand usage at all.

What happens when you hit the limit

Nothing stops. When you exceed included monthly usage, Cursor offers on-demand usage that continues at the same API rates with pay-as-you-go billing settled monthly, or an upgrade to a higher tier. Cursor states that requests are never downgraded in quality or speed when you cross into on-demand.

The Start plan is the exception. It has no on-demand option, so reaching the limit means upgrading to Pro.

Cursor publishes unusually direct spend guidance: daily Tab users and limited agent users typically stay inside included usage, daily agent users typically reach $60 to $100 a month in total usage, and power users running multiple agents often pass $200.

How Cursor's pricing has changed across all these years

Cursor has repriced seven times in 15 months and posted a dated blog for each one.

Date

Milestone

Source

 

28 Jul 2026

Cursor Start launched for India at ₹649/mo covering Grok 4.5 and Composer

cursor.com/blog

1 Jun 2026

Premium Teams seat added for heavy agent users, with real-time usage visibility and spend controls. New customers immediately, renewals from 1 July

cursor.com/blog

11 Feb 2026

Two usage pools introduced, splitting first-party models from third-party API pricing. Composer 1.5 gets 3x Composer 1's limit, briefly 6x

cursor.com/blog

12 Aug 2025

Teams moves from fixed request costs to variable request costs that flex with the work an agent does, effective at the first renewal after 15 September

cursor.com/blog

4 Jul 2025

"Clarifying our pricing". Cursor takes responsibility for communicating the June changes badly and refunds unexpected charges from the prior three weeks

cursor.com/blog

16 Jun 2025

Ultra launched at $200/mo with 20x Pro usage, alongside changes to the Pro plan. Teams unchanged

cursor.com/blog

13 May 2025

MAX moves from per-tool-call pricing to model API pricing

cursor.com/blog

Source: cursor.com/blog, read 22 September 2026.

Flexprice’s Take

Cursor is now one of the more honest pricing pages in AI coding, but only if you find the docs.

Cursor tells you what it charges, which is rarer than it should be. The docs carry a full per-model rate card, and Cursor publishes its own spend guidance: daily agent users typically land between $60 and $100 a month. Almost nobody else in AI coding puts that number in writing.

Metering in tokens helps too. Your bill moves because you picked Claude Fable 5.1 over Composer 2.5, and that's a choice you can see and change.

The catch is that none of this sits on the pricing page. That page sells "3x Pro Agent limits" and never names a unit. The rate card is in the docs, every repricing is in a dated blog post, and comparing Cursor to a rival on pricing pages alone teaches you nothing about what either one charges for.

One thing to model before you commit: the $0.25 per million Cursor Token Rate applies to bring-your-own-key usage, so your own Anthropic key won't get you out of it.

Best For

Teams whose developers understand model cost and will pick accordingly.

Watch Out For

Anyone budgeting from the pricing page instead of the docs, or annualising a plan that has changed seven times in 15 months.

Manish Choudhary

CEO & Co-founder, Flexprice

Building usage-based or credit pricing yourself?

Flexprice handles metering, credits and overage logic so you do not build billing as a second product.

Flexprice’s Take

Cursor is now one of the more honest pricing pages in AI coding, but only if you find the docs.

Cursor tells you what it charges, which is rarer than it should be. The docs carry a full per-model rate card, and Cursor publishes its own spend guidance: daily agent users typically land between $60 and $100 a month. Almost nobody else in AI coding puts that number in writing.

Metering in tokens helps too. Your bill moves because you picked Claude Fable 5.1 over Composer 2.5, and that's a choice you can see and change.

The catch is that none of this sits on the pricing page. That page sells "3x Pro Agent limits" and never names a unit. The rate card is in the docs, every repricing is in a dated blog post, and comparing Cursor to a rival on pricing pages alone teaches you nothing about what either one charges for.

One thing to model before you commit: the $0.25 per million Cursor Token Rate applies to bring-your-own-key usage, so your own Anthropic key won't get you out of it.

Best For

Teams whose developers understand model cost and will pick accordingly.

Watch Out For

Anyone budgeting from the pricing page instead of the docs, or annualising a plan that has changed seven times in 15 months.

Manish Choudhary

CEO & Co-founder, Flexprice

Building usage-based or credit pricing yourself?

Flexprice handles metering, credits and overage logic so you do not build billing as a second product.

Customer
Sentiment Highlights

"Pricing is now a very good deal if you use Grok and Composer"

Cursor user, Hacker News, August 2026

"Since the introduction of the Cursor Token Rate, my company tells us to use Claude and Codex directly"

Engineer at a Cursor customer, Hacker News, August 2026

"I have no idea how much quota I have left until the next block"

Cursor Max subscriber, Hacker News, July 2025

Frequently Asked Questions

Frequently Asked Questions

How much does Cursor cost per month?

Does Cursor charge for overages?

What is the Cursor Token Rate?

Do unused Cursor credits roll over?

Launch usage-based billing this week, not next quarter

Launch usage-based billing this week, not next quarter

Get Instant Feedback on Your Pricing | Join the Flexprice Community with 400+ Builders on Slack

Join the Flexprice Community on Slack