Cursor Pricing: Plans, Usage Pools and Model Costs

Cursor costs $0 on Free, $8 a month on Plus, $30 on Premium, $100 on Super and $200 on Ultra, with Teams seats at $30, $100 or $200. Plans include a weekly pool of usage on Cursor's own models. Claude, GPT and Gemini bill separately at API list price from prepaid credit. Prices checked 2026-10-31.

Prices checked 2026-10-31

Cursor's pricing page and docs changed shape recently. Cursor is now sold as part of The X Subscription, which bundles Grok, Cursor and X. The numbers below come from cursor.com/pricing and the Models and Pricing doc. Prices exclude tax and are subject to change.

The annual column is a saving of two months on each paid tier: $80 against $96, $300 against $360, $1,000 against $1,200 and $2,000 against $2,400. For the Cursor tool itself, see Cursor. For alternatives, see Cursor alternatives.

What changed: legacy plans

Cursor used to sell Hobby, Pro ($20), Pro+ ($60), Ultra ($200) and Teams with Standard ($40) and Premium ($120) seats. Those are now called legacy plans. They keep working as they are, with two monthly usage pools (Cursor Models and Other Models) and on-demand usage billed in arrears. New customers start on the new tiers. Switching is optional and permanent.

Per Cursor's docs, each paid tier includes at least twice as much coding usage on Cursor models as the legacy plan it replaces. It also drops the third-party model allowance. If you are on Pro and use Claude heavily, the new plans change the math, because Claude now bills separately. Cursor says to compare both before switching. One more legacy generation, request-based plans, is older still.

How the new usage model works

One weekly pool

Each paid plan includes a pool of usage for Cursor models. It is shared across Agent in the editor, the CLI, Cloud Agents, Grok Bot and the Grok apps. It refreshes weekly, so one heavy day does not use up your month. Different models draw from it at different rates, and you can see your balance and reset date in editor settings.

Cursor's own models in the pool are Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5. Their rates when you buy extra usage beyond the pool:

Prices per million tokens. Grok 4.7 context above 256K input tokens is billed at 2x standard rates up to 500K. Fast mode is 2x.

Third-party models

Claude, GPT and Gemini are not in the pool. On Premium and above you can use them at the model's API list price, from the first request, paid from prepaid credit. Examples from the doc's table, per million tokens: Claude Opus 5.5 is $4 in and $20 out, Claude Sonnet 5.5 is $2 and $10, and Claude Haiku 5.5 is $0.10 and $0.50. These match Anthropic's own prices, which you can check on Claude API pricing. On Teams and Enterprise, eligible third-party requests add a Cursor Token Rate of $0.25 per million tokens on top, applying to input, output and cached tokens. Individual plans do not pay it, and it does not apply to Cursor's own models.

Top-ups and spending caps

Prepaid credit pays for third-party models and for Cursor models beyond your weekly pool. You buy credit when you choose or turn on Auto Top-up to refill below a threshold. A monthly additional usage limit caps paid usage regardless of how much credit you hold. On Teams, admins manage a shared balance and one cap for the team.

When the weekly pool runs out you can wait for the reset or keep working on prepaid credit at the same per-model rates.

What each plan includes

Plus is deliberately narrow. Per the docs it gives Agent with Cursor's own models at fixed settings (Fast off, effort medium, and Grok 4.7 on the 256K window), plus Grok Chat, Voice, Imagine, Build and Grok API access. It has no Auto, no Cloud Agents, no Automations, no Bugbot, no Origin, no third-party models and no bring-your-own-key.

Cursor recommends Premium for new AI users, Super for daily agent users and Ultra for power users running several agents or automations. It recommends Plus for social creators with limited AI usage, which tells you who the tier is aimed at.

Teams and Enterprise

Teams seats are $30, $100 or $200 a month and match the individual tier of the same usage. Teams adds centralized billing and administration, usage analytics, team-wide privacy settings, custom chat data retention, SAML and OIDC single sign-on, a team marketplace for rules, skills and plugins, and mix-and-match seat types.

Enterprise adds invoice and PO billing, SCIM, granular spend controls, group-level settings and limits, auto-run, browser and network controls, repository, model and MCP access controls, audit logs and service accounts, and priority support. Usage is pooled and the price is custom.

Privacy

Cursor says privacy mode can be turned on in settings or by a team admin, and when it is on, "we guarantee that code data is not used for training by us or our model providers." Third-party providers are subject to their own terms, and some premium models require data-retention approval for Teams and individual customers.

Buying safely

Cursor states subscriptions are sold only through X Subscription Corp, directly from its official website, and it does not authorize resellers. It says accounts bought elsewhere may be suspended. Cancel at least 24 hours before renewal to avoid another charge.

Cursor against other tools

The difference that matters is the model. Cursor Plus and the included pool use Cursor's own models. If you want Claude inside Cursor, you pay Anthropic's list rates from credit. Claude Code, by contrast, gives Claude through a flat plan with a usage limit; see Claude Code pricing. The AI coding agent pricing comparison puts them in one table.

A worked monthly example

This is arithmetic from published rates, not a measurement of anyone's usage. Suppose a developer runs agent sessions on Claude Sonnet 5.5 through Cursor Premium, and a typical working day sends 2 million input tokens and 100,000 output tokens in total, ignoring caching.

With prompt caching, the input cost falls, and Cursor's table lists a $0.10 cache read for Sonnet 5.5, so a long-running session with a stable prefix costs less. Switch the same work to Opus 5.5 at $4 and $20 and the daily figure doubles to $10. The comparison with a flat plan is useful: Claude Max is $100 a month, with 5x Pro usage, for Claude models inside Claude Code only. If your agent use is heavy and Claude-centric, the flat plan may be cheaper. If you switch between models or work mostly in an editor, Cursor's metered model is more flexible. Check Claude Max plan for what that plan's limits mean.

Who should pick which

Our take

Cursor's price list is simple to read and hard to forecast. The plan fee covers Cursor's own models. Everything else, the Claude and GPT models most people think of, bills as extra from prepaid credit at list price. That makes Cursor closer to "editor plus metered models" than to "all-you-can-use".

Two practical consequences. First, set the monthly additional usage limit on day one. Without it, a heavy Agent session on Opus 5.5 at $4 and $20 per million tokens can spend credit faster than you expect. Second, compare against Claude Code directly if Claude is your main model: Claude Code on a $17 or $100 plan has a ceiling, while Cursor on credit has none.

We did not test either tool, so we have no view on speed or quality. On price structure alone, Cursor suits people who want one editor with many models and accept variable bills. Fixed-bill buyers should look at the flat plans in GitHub Copilot pricing and Claude Pro pricing.

Checklist

• Decide whether you need third-party models. If yes, Premium or above.
• Set a monthly additional usage limit before the first top-up.
• Turn on privacy mode, or ask your admin to.
• Buy only from cursor.com.
• If on a legacy plan, compare before switching, because switching is permanent.
• Recheck the page. Plans and rates have changed recently.

Questions we hear

Is Cursor still a standalone product? Cursor is sold as part of The X Subscription, one subscription for Grok, Cursor and X, per Cursor's own docs. The editor and agent features are still what the Cursor plans cover, and the pricing page lists Cursor features beside Grok and X benefits.

Do I have to pay for X features I do not want? The plans bundle them. The pricing page lists X benefits such as the Premium checkmark and ad-free scrolling on every paid tier. If you want only the editor, you pay the bundle price.

Can I bring my own API key? On Plus, no, per the docs. For other tiers the page lists third-party models at API list price from prepaid credit, and the Cursor Token Rate applies on Teams and Enterprise to bring-your-own-key usage too.

What does Auto do? Plus has no Auto. On Teams and Enterprise a router picks between Cursor and third-party models per request by optimization mode, and a third-party route bills as third-party usage.

When does the weekly pool reset? Weekly. The next reset date is shown in editor settings and on the Grok dashboard.

Planning a team rollout

Start with Premium seats at $30 for everyone who will use the agent daily, and set one team-wide additional-usage cap. Then watch the admin dashboard for a month. Move people who run several agents or automations to $100 or $200 seats, and leave occasional users on the lowest tier. Cursor's mix-and-match seat types exist for exactly this. Turn on team-wide privacy settings first, since privacy mode is the control that guarantees code is not used for training by Cursor or its model providers.