Global AI Token Comparison
Global AI Token Plan comparison home

Cursor Token API Pricing, Top-up, Deals & Official Entry

This Cursor API page focuses on API key access, pricing comparison and subscription-linked deals: Grok 4.6 is the newest first-party flagship at $2 input, $6 output and $0.5 cache read per 1M tokens, with a 50% launch-week discount effective 2026-08-12; Grok 4.5 matches at $2/$6, Composer 2.5 is the cheapest at $0.5/$2.5, and Auto routes models at $1.25 input + cache write, $6 output and $0.25 cache read per 1M tokens. It is not a standalone OpenAI-compatible endpoint; calls go through Cursor IDE, Cloud Agents, CLI or Cursor SDK with CURSOR_API_KEY from Dashboard Integrations, billed alongside the Auto + Composer pool, API pool, Teams Cursor Token Rate at $0.25/M, on-demand arrears, Max Mode and rate-limit boundaries.

Token APIGrok 4.6 · SDK
SubscriptionToken API

Cursor Token API Latest Updates

Model

Cursor Grok 4.6 Launch: 50% Off First Week, $2/$6 Pricing Comparison vs Composer 2.5

Cursor's newest first-party flagship Grok 4.6 is live at $2 input, $6 output and $0.5 cache read per 1M tokens, with 50% off during its launch week from 2026-08-12. This pricing comparison breaks down how Grok 4.6 sits against Grok 4.5, Composer 2.5 and Auto, benchmarks it against Claude Sonnet 5 and Haiku 4.5, and ties in Token Plan Pro/Pro+/Ultra quota, API key setup and the official entry point so you can judge whether Grok 4.6 is worth switching to.

View all updates

Below is the complete pricing comparison for Cursor Token API, covering 12 plan tiers, core model capabilities, quotas, and official entry. All data is sourced from the official website to help you decide whether it’s worth it.

Last updated：

Cursor Token API Price Comparison: Core Models

Grok 4.6Grok 4.6 FastGrok 4.5Grok 4.5 FastComposer 2.5Composer 2.5 FastAuto CostAuto BalanceAuto IntelligenceComposer 2Composer 1.5Composer 1
Grok 4.6

Cursor's newest first-party flagship at $2/$6 per MTok (50% off launch week from 2026-08-12); Grok 4.6 Fast at $4/$12 covers low-latency needs.

Composer 2.5

The low-cost agentic coding model in Cursor's pool at $0.5/$2.5 per MTok—the public IDE default and recommended SDK model id for daily and cloud agents.

Composer 2

Prior Composer generation at the same $0.5/$2.5 rates—hidden by default for legacy SDK integrations or comparison tests.

Composer 1.5

Legacy high-tier Composer at $3.5/$17.5 per MTok—hidden by default and costlier than 2.x, mainly for compatibility.

Composer 1

First-generation Cursor agent model at $1.25/$10 per MTok—hidden by default; prefer 2.5 for new work.

Cursor Token API Price Comparison: Plans

Grok 4.6
Newest Cursor flagshipRecommended
Input
$2
Output
$6
Cache Read $0.5/M · 50% off launch week (from 2026-08-12)
Usage
Grok 4.6 is Cursor's newest first-party flagship at $2 input, $0.5 cache read, and $6 output per 1M tokens; a 50% launch-week discount applies from 2026-08-12 per the official pricing page.
Best for
Agent development needing the newest first-party flagship
Grok 4.6 Fast
Newest Cursor flagship Fast
Input
$4
Output
$12
Cache Read $1/M
Usage
Grok 4.6 Fast pricing is $4 input, $1 cache read, and $12 output per 1M tokens.
Best for
Newest flagship users prioritizing low latency
Grok 4.5
Previous flagship
Input
$2
Output
$6
Cache Read $0.5/M · first-party
Usage
Grok 4.5 is the previous first-party flagship at $2 input, $0.5 cache read, and $6 output per 1M tokens—same rates as Grok 4.6 but not the newest.
Best for
Users preferring a proven, stable first-party model
Grok 4.5 Fast
Previous flagship Fast
Input
$4
Output
$12
Cache Read $1/M
Usage
Grok 4.5 Fast pricing is $4 input, $1 cache read, and $12 output per 1M tokens.
Best for
Previous flagship users prioritizing low latency
Composer 2.5
Budget mainlineRecommended
Input
$0.5
Output
$2.5
Cache Read $0.2/M · public IDE default · SDK model id: composer-2.5
Usage
Composer 2.5 is the low-cost agentic coding model in Cursor's pool at $0.5 input and $2.5 output per 1M tokens with $0.2/M cache reads—for daily agents, cloud agents, and Cursor SDK automation.
Models
Manual Composer 2.5 draws the Token Plan API pool; Auto or Composer 2.5 via Auto routing draws the Auto + Composer pool with typically more included usage.
Highlights
Cursor SDK public beta supports composer-2.5 in TypeScript (@cursor/sdk) and Python (cursor-sdk)—same token rates as the IDE under standard token-based consumption pricing against subscription or on-demand.
Best for
IDE and SDK users who want Cursor's own agent model
Composer 2.5 Fast
Fast
Input
$3
Output
$15
Cache Read $0.5/M
Usage
Composer 2.5 Fast is $3 input, $0.5 cache read, and $15 output per 1M tokens.
Best for
Composer users prioritizing low latency
Auto Balance
Balanced routing
Price Comparison
Actual model API price
Token Rate applies when routed to third parties
Usage
Billed at the actual routed model API price.
Best for
Balancing capability and cost
Auto Intelligence
Intelligence-first routing
Price Comparison
Actual model API price
Token Rate applies when routed to third parties
Usage
Billed at the actual routed model API price.
Best for
Prioritizing model intelligence
Composer 2
Prior generation
Input
$0.5
Output
$2.5
Hidden by default · Cache Read $0.2/M
Usage
Composer 2 matches 2.5 pricing at $0.5 input and $2.5 output per 1M tokens—hidden by default with new users routed to Composer 2.5.
Models
You can still pick composer-2 in the model selector or SDK for legacy workflows—same billing as 2.5 from the API or Auto + Composer pool.
Best for
Legacy integrations still on Composer 2
Composer 1.5
Legacy high tier
Input
$3.5
Output
$17.5
Hidden by default · Cache Read $0.35/M
Usage
Composer 1.5 is priced at $3.5 input and $17.5 output per 1M tokens with $0.35/M cache reads—above Composer 2.x and hidden by default; not recommended for new work.
Models
If manually selected, it bills at API pool token rates—for rare compatibility cases that must stay on legacy Composer behavior.
Best for
Compatibility cases pinned to Composer 1.5
Composer 1
First generation
Input
$1.25
Output
$10
Hidden by default · Cache Read $0.125/M
Usage
Composer 1 is priced at $1.25 input and $10 output per 1M tokens with $0.125/M cache reads—Cursor's first agent model, hidden by default.
Models
New development should use Composer 2.5; pick Composer 1 only for historical projects or comparison tests.
Best for
Historical projects or comparison tests
Auto
Smart routing
Input
$1.25
Output
$6
Input+Cache Write $1.25/M · Cache Read $0.25/M · Auto+Composer pool
Usage
Auto lets Cursor pick models balancing intelligence, cost, and reliability—official Auto pricing: Input+Cache Write $1.25, Output $6, Cache Read $0.25 per 1M tokens.
Models
Choosing Auto or Composer 2.5 via Auto routing draws the Auto + Composer pool with generous included usage—typically more allowance-friendly than manually picking third-party flagships.
Highlights
Auto is exempt from the Teams Cursor Token Rate $0.25/M; non-Auto agent requests add that surcharge on top of model API pricing.
Best for
Daily development where Cursor picks models and controls cost

Cursor Token API Price Comparison: Notes

  • Cursor does not expose a standalone OpenAI-compatible REST inference endpoint; Composer models are callable only via the Cursor IDE, CLI, Cloud Agents, and Cursor SDK. Generate API keys on the Cursor Dashboard Integrations page.
  • Official pricing table Notes: Composer 1/1.5/2 are hidden by default; the Cursor models pool contains Grok 4.6, Grok 4.5, and Composer 2.5, with Grok 4.6 at 50% off during its launch week from 2026-08-12.
  • Teams/Enterprise third-party models and Auto Balance/Intelligence routes to third parties add the $0.25/M Cursor Token Rate; Auto Cost and all first-party Cursor models (Grok 4.6, Grok 4.5, Composer 2.5) are exempt. Max Mode is legacy request-based only at model API price plus 20%.
  • After included API usage, on-demand continues at the same token rates billed in arrears; official docs state requests are never downgraded in quality or speed.

Cursor Token API Price Comparison: Tools & Integration

Cursor SDK (TypeScript)cursor-sdk (Python)CURSOR_API_KEYCloud Agents APICursor IDE

Cursor Token API Pricing, Top-up, Deals, Quota, Usage, Setup & Updates

AnalysisPublished Updated

Cursor API Pricing Comparison: Grok 4.6 vs Composer 2.5 vs Auto Costs, API Key Setup and Is It Worth It

CursorAPIPricing ComparisonGrok 4.6Grok 4.5Composer 2.5

Cursor API pricing comes down to three lines: the newest flagship Grok 4.6 at $2 input / $6 output per 1M tokens (50% off launch week from 2026-08-12), Composer 2.5 at $0.5/$2.5 as the cheapest, and Auto routing at $1.25 input + cache write / $6 output; the previous Grok 4.5 matches Grok 4.6 at $2/$6. Cursor API is not a standalone OpenAI-compatible endpoint—it bills through your Cursor subscription's API or Auto + Composer pool via CURSOR_API_KEY from Dashboard Integrations. This guide covers the rate table, caching and usage pools, API key setup for IDE/CLI/Cloud Agents/SDK, rate limits, and whether it's worth pairing with a Cursor subscription.

View more updates

Page published：

Cursor Token API Pricing, Top-up, Deals & FAQ

AI Token Plan

Global AI Token Comparison

AI Token Plan is not just a collection of vendor links. It places 41+ domestic and international AI platforms into one comparison framework, covering 395+ model entries plus common plan types such as Token Plans, Coding Plans, IDE Tools and LLM APIs.

When you need to compare AI subscription pricing, coding plan quota, API usage cost or official deal entry points, AI Token Plan brings official prices, plan tiers, usage rules, model capabilities and tool integrations into one place, reducing the need to check multiple vendor sites manually.

Data is continuously organized as vendor pricing pages, plan pages and product documentation change, making the homepage a pricing comparison entry point while detail pages explain whether each vendor plan fits individual developers, team purchasing or long-term API usage.

Prices come from each platform's official site and may change at any time; the official price prevails.

© 2026 AI Token Plan · All rights reserved · First published June 18, 2026 · 64 days running · Sitemap