Short answer: Cursor’s documentation lists individual plans from Free to $200/month. Paid plans include a weekly pool for Cursor models, while third-party models are billed separately inside Cursor at the listed rates. Below are representative model rates per million tokens; these charges are distinct from the subscription fee and do not establish what a separate production API will cost. Cursor displays prices with a dollar sign and says plan prices exclude tax; its pricing page does not specify currency or regional applicability in the cited pricing details. (Cursor Models & Pricing)
Cursor plan prices
Cursor’s pricing documentation lists these plan prices and included-usage tiers. Annual billing is listed for Plus through Ultra. Prices exclude tax. The page does not state currency or regional applicability in the cited pricing details, so treat these as listed figures—not a confirmed quote for a particular location. (Cursor Models & Pricing)
| Plan | Listed price | Included usage and notable differences |
|---|---|---|
| Free | $0 | Limited Agent and Chat usage; third-party models unavailable |
| Plus | $8/month or $80/year | Base weekly pool; Cursor models at fixed settings; no third-party models |
| Premium | $30/month or $300/year | 4× Plus usage; third-party models at API list price; adds Cloud Agents and Automations |
| Super | $100/month or $1,000/year | 15× Plus usage; priority model access at peak times |
| Ultra | $200/month or $2,000/year | 50× Plus usage; earliest access to new products and models |
| Teams | $30, $100, or $200 per seat/month | Premium-, Super-, or Ultra-level seats; tiers can be mixed; no Plus seats |
| Enterprise | Custom | Pooled usage and additional enterprise controls |
The annual figures are annual-billing prices, not monthly rates. Dividing by 12 gives approximate monthly equivalents of $6.67 for Plus, $25 for Premium, $83.33 for Super, and $166.67 for Ultra. The included usage is not a fixed number of requests: each paid plan has a weekly pool, and different Cursor models draw from it at different rates. The pool is shared across Cursor and covered Grok usage on the subscription. (Cursor Models & Pricing)
Choosing a tier: Plus is the lower-cost option if its included Cursor models and fixed settings are sufficient. Premium or higher is required for third-party models inside Cursor. Super and Ultra increase the included pool and add priority or early access; compare those benefits with your usage rather than assuming a plan buys a fixed number of prompts. (Cursor Models & Pricing)
Legacy Cursor plans are different
Cursor says existing customers on legacy plans—including Pro and Pro+—can keep those plans. The documentation describes legacy billing as two monthly usage pools, for Cursor Models and Other Models, with on-demand usage instead of the weekly-pool and top-up arrangement described for the plans above. The pricing information cited here does not establish legacy Pro or Pro+ prices, so this article does not assign them dollar amounts. (Cursor Models & Pricing)
Model rates and extra charges inside Cursor
Cursor’s model pages list rates per million tokens. These are charges for using the named third-party models through Cursor, billed from prepaid credit on The X Subscription; the pages say third-party models are not part of the plan’s included Cursor-model pool. The table shows standard rates for representative Claude models: (Claude Haiku 5.5, Claude Sonnet 5.5, Claude Opus 5.5)
| Model | Input | Cache write | Cache read | Output |
|---|---|---|---|---|
| Claude Haiku 5.5, up to 100k input tokens | $0.10 | $0.125 | $0.01 | $0.50 |
| Claude Sonnet 5.5, standard rate | $2 | $2.50 | $0.10 | $10 |
| Claude Opus 5.5, standard rate | $4 | $5 | $0.20 | $20 |
All amounts are dollars per million tokens as displayed in Cursor’s documentation. Haiku 5.5 requests above 100k input tokens are listed at five times the standard rates: $0.50 input, $0.625 cache write, $0.05 cache read, and $2.50 output per million tokens. Opting into regional data residency adds a 10% uplift to model pricing for eligible models; confirm whether a specific model and region qualify before estimating spend. (Claude Haiku 5.5, Cursor Models & Pricing)
For a separate example of a Cursor model’s paid overage, Cursor lists Grok 4.5 additional usage beyond the weekly pool at $2 per million input tokens and $6 per million output tokens at standard speed; its Fast variant is listed at $4 and $18, respectively. These are overage rates, not the third-party model rates above. (Grok 4.5)
What happens when included usage runs out?
- Cursor models: Their usage draws from the weekly pool. When it runs out, Cursor lists waiting for the reset, using prepaid credit for additional usage, or upgrading as options. A monthly limit can cap paid additional usage regardless of the credit balance. (Cursor Models & Pricing)
- Third-party models in Cursor: Free and Plus do not include them. On Premium and above, Cursor bills them at the model’s API list price from prepaid credit, starting with the first request. This is an in-Cursor charge path, not a separate production API quote. (Cursor Models & Pricing)
- Teams and Enterprise: The $0.25-per-million-token Cursor Token Rate applies to eligible third-party requests selected directly or routed through Balance or Intelligence. It is charged on input, output, and cached tokens, in addition to model charges. Cursor says the fee does not apply to Cursor models, Cost-mode requests, or individual plans; eligible third-party models with cache-read prices strictly below $0.10 per million tokens across every pricing tier are also exempt. (Cursor Models & Pricing)
Routing affects which budget is used: Cost mode bills as a Cursor-model request and draws from the weekly pool, regardless of which model serves it. On Teams and Enterprise, Balance or Intelligence routes draw from the pool when Cursor serves the request with a Cursor model. If the router selects a third-party model, the request is billed at that model’s list price from prepaid credit and may also incur the Token Rate, subject to the exemption above. (Cursor Models & Pricing)
Cursor billing is not production API pricing
Cursor’s pricing documentation describes subscription use and model charges inside Cursor. It does not establish the price, terms, or suitability of a separately used production API for an application. Those depend on the provider, model, workload, and applicable service terms; no production provider or workload is specified here.
Cursor also documents a bring-your-own-API-key option for supported chat models. In that arrangement, the model provider bills for model usage; eligible Teams and Enterprise requests may still incur Cursor’s fee. That option is not the same as integrating and budgeting for an API directly in a production application. (Cursor’s API-key documentation)
Budget takeaway: Consider the Cursor subscription, any paid use beyond included pools, third-party model charges within Cursor, and any applicable team Token Rate separately. For a production API budget, use the chosen provider’s current pricing and terms against a representative workload. A Cursor plan price or in-Cursor model rate alone does not answer that separate cost question.
Sources
Models & Pricing | Cursor Docs
Claude Haiku 5.5 | Cursor Docs