Cursor Token Rate: 15 models skip the $0.25 per million charge on Teams and Enterprise

Who pays the Cursor Token Rate

Cursor's Models & Pricing page, checked October 22, 2026, says eligible third-party model requests on Teams and Enterprise plans include a Cursor Token Rate of $0.25 per million tokens. It is charged on top of the model's API price, for both additional usage and BYOK usage, and it applies to input tokens, output tokens and cached tokens. Individual plans don't pay it.

The rate applies when you pick an eligible third-party model directly. It also applies when Cursor Router sends a Balance or Intelligence request to one. It does not apply to Cursor models, including Grok and Composer, or to Cost mode requests. The page gives no effective date for the rate or its exclusions.

Sources: Models & Pricing

The 15 models excluded from the rate

The page lists 15 third-party models that skip the Cursor Token Rate: Claude Haiku 5.5, Gemini 2.5 Flash, Gemini 3 Flash, Gemini 3.1 Flash Lite, Gemini 3.5 Flash Lite, Gemini 3.7 Flash, Gemini 3.8 Flash, GLM 5.3 Flash, GPT-4.1 Nano, GPT-4o Mini, GPT-5 Mini, GPT-5.1 Codex Mini, GPT-5.4 Mini, GPT-5.4 Nano and GPT-5.6 Luna.

In the plans table, third-party usage on both Teams and Enterprise is priced at API list price plus the Cursor Token Rate on eligible models.

Sources: Models & Pricing

What it means for a Teams or Enterprise bill

Third-party models such as Claude, GPT and Gemini are not part of the weekly pool. These charges come out of prepaid credit, which Teams admins manage as one shared balance with one cap for the whole team. A request to a model that is not excluded pays the model's list price plus $0.25 per million tokens. For example, Claude Sonnet 5.5 is listed at $2 per million input tokens and $10 per million output tokens and carries the extra rate. Claude Haiku 5.5, at $0.10 input and $0.50 output, does not.

Cost mode, the classic Auto, bills as a Cursor-model request and draws from the weekly pool whichever model serves it, with no Token Rate. A Balance or Intelligence request that the router sends to a third-party model is paid from prepaid credit at that model's list price. It also pays the Token Rate unless the model is on the excluded list. A monthly additional usage limit caps paid additional usage, however much credit you hold. You manage prepaid credit, Auto Top-up and that limit under Additional usage credits at grok.com/dashboard/billing.

Sources: Models & Pricing

Data residency uplift on the same page

The page also says opting in to regional data residency adds a 10% uplift to model pricing for eligible models. Its pricing tables mark US data residency as Yes for the Grok, Composer, Claude and GPT-5.6 models shown. They mark it as No for Gemini 3.1 Pro, Gemini 3.8 Flash, Gemini 4 Argon and Muse Wonder.

Sources: Models & Pricing

Corrections

2026-10-22: Updated factual details after a change in the linked official source.