Home / Guides / Cursor model pricing
Cursor model pricing, model by model
Cursor prices every model per million tokens, and what you pay depends on whose model it is. Cursor's own models, Grok 4.7, 4.6 and 4.5 and Composer 2.5, come out of the weekly usage pool on every paid plan; past the pool they bill at Cursor's rates, from $0.50 input and $2.50 output on Composer 2.5 to $2 and $6 on Grok 4.7. Claude, GPT and Gemini are not in the pool. On Premium and above you pay their API list price from prepaid credit from the first request, $4 and $20 on Claude Opus 5.5 or $2 and $12 on Gemini 3.1 Pro, and on Plus and Free you cannot use them. Teams and Enterprise add $0.25 per million tokens on most third-party models.
Checked 1 November 2026 against Cursor's Models & Pricing page, its usage and limits help and its legacy plans page · By Soren Starck
SessionWatcherSee your Cursor spend
without opening the dashboard.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
On Windows? Cursor usage in the Windows taskbar: the same spend and budget readout on Windows 10 and 11.
How much do Cursor's own models cost?
Grok and Composer are the only models in the weekly pool, which Cursor shares with Grok Chat, Grok Bot and the other Grok apps on The X Subscription. Each model draws from the pool at its own rate, and once the pool is used up the rates below come out of your prepaid credit. All prices are per million tokens; none of these models charges for cache writes.
| Cursor model | Input | Cache read | Output |
|---|---|---|---|
| Grok 4.7 | $2 | $0.50 | $6 |
| Grok 4.7 (Fast) | $4 | $1 | $12 |
| Grok 4.7 500k | $4 | $1 | $12 |
| Grok 4.7 500k (Fast) | $6 | $1.50 | $18 |
| Grok 4.6 | $2 | $0.50 | $6 |
| Grok 4.6 (Fast) | $4 | $1 | $12 |
| Grok 4.5 | $2 | $0.50 | $6 |
| Grok 4.5 (Fast) | $4 | $1 | $18 |
| Composer 2.5 | $0.50 | $0.20 | $2.50 |
| Composer 2.5 (Fast) | $3 | $0.50 | $15 |
Grok 4.7 bills long context (more than 256k input tokens) at twice these rates, up to 500k, and fast mode on long context at three times. Fast mode costs twice the standard rate on Grok 4.7 and 4.6 (three times the output price on Grok 4.5) and six times the input and output price on Composer 2.5. Prices from Cursor's table, US dollars, before the 10% that regional data residency adds.
How much does Claude, GPT or Gemini cost in Cursor?
Third-party models bill at the provider's API list price from your first request, paid from prepaid credit, on Premium ($30 a month), Super and Ultra. Plus ($8) and Free include Cursor's own models only. These are the 11 third-party models Cursor shows by default; its table lists 42 more that are hidden in the model picker, older Claude, GPT-5 and Gemini versions, DeepSeek, GLM and Kimi among them.
| Model | Input | Cache write | Cache read | Output |
|---|---|---|---|---|
| Claude Fable 5.1 | $10 | $12.50 | $0.25 | $50 |
| Claude Opus 5.5 | $4 | $5 | $0.20 | $20 |
| Claude Sonnet 5.5 | $2 | $2.50 | $0.10 | $10 |
| Claude Haiku 5.5 | $0.10 | $0.125 | $0.01 | $0.50 |
| GPT-5.6 Sol | $4 | $5 | $0.40 | $20 |
| GPT-5.6 Terra | $2 | $2.50 | $0.20 | $12 |
| GPT-5.6 Luna | $0.20 | $0.25 | $0.02 | $1.20 |
| Gemini 3.1 Pro | $2 | None | $0.20 | $12 |
| Gemini 4 Argon | $2 | None | $0.10 | $10 |
| Gemini 3.8 Flash | $0.75 | None | $0.075 | $3.75 |
| Muse Wonder | $4 | None | $0.20 | $20 |
Per million tokens. Claude Haiku 5.5 bills five times these rates above 100k input tokens. GPT-5.6 models charge twice the input price for long context, up to 1M tokens, and twice the rates in fast mode. Claude Sonnet 5.5 keeps the same rates up to 1M tokens.
What does a task cost in Cursor?
Cursor bills tokens, not tasks, so the honest answer is arithmetic. Take a mid-sized agent task that sends 50k new input tokens, reads 350k from the prompt cache as the conversation grows, and writes 20k tokens of code and replies. The cached tokens outnumber the new ones seven to one, so the cache read price weighs as much as the headline input price. At Cursor's rates:
| Model | The same task |
|---|---|
| Composer 2.5 | $0.15 |
| Grok 4.7 | $0.40 |
| Claude Sonnet 5.5 | $0.34 |
| Gemini 3.1 Pro | $0.41 |
| Claude Opus 5.5 | $0.67 |
| GPT-5.6 Sol | $0.74 |
Our arithmetic on Cursor's per-token prices, not a Cursor figure; the token counts are an example. Cache writes, which the Claude and GPT models charge when a prompt is first stored, would add to their figures. On a paid plan the Grok and Composer rows come out of the weekly pool first.
Which Cursor model is cheapest?
Among Cursor's own models, Composer 2.5 at $0.50 input and $2.50 output per million tokens, a quarter of Grok 4.7's input price. Among the third-party models shown by default, Claude Haiku 5.5 at $0.10 and $0.50 for requests up to 100k input tokens, then GPT-5.6 Luna at $0.20 and $1.20. The most expensive shown by default is Claude Fable 5.1 at $10 and $50, a hundred times Haiku 5.5. Cheap per token is not cheap per task if a model needs more turns to finish, so compare on a week of your own work.
When do you pay per token in Cursor?
Three cases, on The X Subscription plans. Third-party models always, from the first request. Cursor's own models once the weekly pool is used up, if you keep going instead of waiting for the reset or upgrading. And on Teams and Enterprise, the Cursor Token Rate of $0.25 per million tokens on eligible third-party requests, your own API keys included. Credit comes from top-ups or Auto Top-up, a monthly additional usage limit caps what it can spend, and each top-up expires 365 days after purchase. Requests are not downgraded when the pool runs out.
The legacy plans (Hobby, Start, Pro, Pro+, Ultra and the old Teams seats) work differently: two monthly pools, Cursor Models and Other Models, with third-party models drawn from Other Models at their API price, and on-demand usage past them. Max Mode on the legacy request-based plans adds 20% to a model's API rate. How those plans compare is in Cursor Pro vs Pro+ vs Ultra, and what each current plan costs a month is in Cursor pricing.
Seeing what each model cost you
On The X Subscription, Usage at grok.com/dashboard shows how much of the weekly pool you have used, the next reset and your prepaid credit, and billing lives at grok.com/dashboard/billing. On a legacy plan, the Spending tab at cursor.com/dashboard shows both pools, the allowance left and any on-demand charges. How the limits themselves work is in Cursor rate limits explained.
On a legacy plan, SessionWatcher keeps that spend, the budget left and the cycle's timing in the Mac menu bar or the Windows taskbar, and notifies you at 70% and 90% by default, so a run on an expensive model shows up while it is happening rather than on the next bill. Every other way to read it is in how to check Cursor usage.
SessionWatcherWatch your Cursor spend
before the expensive model eats it.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
Frequently asked questions
How much does Claude cost in Cursor?
Claude is a third-party model, so it is not in the weekly pool. On Premium, Super and Ultra you pay Anthropic’s API list price from prepaid credit from the first request: per million tokens, Claude Opus 5.5 is $4 input and $20 output, Claude Sonnet 5.5 $2 and $10, Claude Haiku 5.5 $0.10 and $0.50, and Claude Fable 5.1 $10 and $50. Plus and Free cannot use Claude. Teams and Enterprise add the Cursor Token Rate of $0.25 per million tokens on most third-party models.
Are Grok and Composer free in Cursor?
They are included, not free. Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5 draw from the weekly usage pool on every paid plan, and different models draw from it at different rates. Once the pool is used up, the same models bill at Cursor’s rates from prepaid credit, for example $0.50 input and $2.50 output per million tokens on Composer 2.5. The Free plan has limited usage.
What is the Cursor Token Rate?
A charge of $0.25 per million tokens on eligible third-party model requests on Teams and Enterprise plans, on input, output and cached tokens, on top of the model’s API price. It applies to your own API keys too. It does not apply to Cursor’s own models or to individual plans, and some low-cost models are excluded, Claude Haiku 5.5, GPT-5.6 Luna and Gemini 3.8 Flash among them.
How much does Max Mode cost in Cursor?
Max Mode exists only on the legacy request-based plans. It extends a model’s context window beyond the default and bills at the model’s API rate plus 20%. On The X Subscription plans, long context is priced per model instead: Grok 4.7 at twice its rates above 256k input tokens, GPT-5.6 at twice the input price up to 1M tokens, Claude Sonnet 5.5 up to 1M tokens at its normal rates.
Where do I see what I spent on models in Cursor?
On The X Subscription plans, under Usage at grok.com/dashboard, which shows the weekly pool used, the next reset and your prepaid credit; your editor settings show the same remaining usage. Billing, top-ups and the monthly spending cap are at grok.com/dashboard/billing. On a legacy plan, the Spending tab at cursor.com/dashboard shows both pools, the allowance left and any on-demand charges.
Sources
All read on 1 November 2026:
- Cursor: Models & Pricing: the plans, the weekly pool, every model's price per million tokens, long context and fast mode, data residency, top-ups, the Cursor Token Rate and Max Mode.
- Cursor: Usage and limits: where each plan shows usage, the resets, prepaid credit expiry and your own API keys on Teams.
- Cursor: Legacy Cursor plans: the two monthly pools and on-demand usage.