console only units unstated
As advertised. RESTRUCTURED 2026-08-28: no dollar figure published any more. Two separate pools per plan, both reset monthly: "Cursor Models" (Grok 4.7, Grok 4.6, Grok 4.5, Composer 2.5 -- "significantly more included usage") and "Other Models" (third-party, billed at API rate against the included pool). Pro/Pro+/Ultra include both pools; the India-only Start plan (₹649/mo) includes only the Cursor Models pool.
What that actually means. Until 2026-08-28 this was dollars at model API rates ($20/$70/$400 included, Pro ~225 Sonnet requests). The docs page now only says each pool is "included" per plan, with no dollar or token figure -- a real disclosure regression from before. Rough guidance offered instead: daily Tab users and limited Agent users "typically stay within included usage"; daily Agent users run $60-100/mo total; power users with multiple agents run $200+/mo.
Where to check what is left. cursor.com/dashboard -> Usage
Real multipliers that remain: cache WRITES 1.25x input. Long context costs 2x input on the GPT models (GPT-5.4/5.5/5.6) and, since 2026-09-24, on Grok 4.7 too, which has a 500k variant at $4/$12 against its own $2/$6. The rest reach 1M "at the same per-token rates". On Teams/Enterprise a separate "Cursor Token Rate" of $0.25/M applies on top of API price for third-party models (first-party Cursor models are exempt). Opting into regional data residency adds a 10% uplift on model pricing. Max Mode survives only on legacy request-based plans, billed at the model's API rate plus 20%. New: Cursor Router gives Auto three modes (Cost, Balance, Intelligence) on Teams/Enterprise. 2026-09-24: Claude Opus 5.5 se vende ya en el pool Other Models a precio de API ($4 entrada, $5 escritura de cache, $0,2 lectura, $20 salida), junto con Fable 5.1, Sonnet 5, Gemini 3.1 Pro, Gemini 3.8 Flash y Muse Spark 1.3. Ningun GPT-6: el mas alto sigue siendo GPT-5.6 Sol. OJO: la tabla de terceros esta truncada tras un boton, asi que ese listado es parcial. 2026-10-02: Claude Sonnet 5.5 entra en el pool de terceros a precio de API ($2 entrada, $2,50 escritura de cache, $0,20 lectura, $10 salida) y Sonnet 5 desaparece de la tabla. Precios exactos de los demas: Fable 5.1 $10/$12,5/$0,25/$50, Opus 5.5 $4/$5/$0,2/$20, Gemini 3.1 Pro $2/-/$0,2/$12, Gemini 3.8 Flash $0,75/-/$0,075/$3,5, Muse Spark 1.3 $1,25/-/$0,15/$4,25, GPT-5.6 Sol $4/$5/$0,4/$20, Terra $2/$2,5/$0,2/$12 y Luna $0,2/$0,25/$0,02/$1,2. GPT-6.1 Sol NO esta: su tope sigue siendo GPT-5.6 Sol. Y OJO: la regla del 2x por contexto largo en los GPT ya no aparece en esta pagina, asi que pasa a no confirmada; lo que hay son filas de precio aparte, como el Grok 4.7 de 500k.
checked 2026-10-02 · high (price) / none (quota -- vendor stopped publishing a number) · source