Cursor AI pricing in 2026: Plans from $0 to $200, models, limits and alternatives

Plans, prices and features checked on the official pages on October 21, 2026.

Cursor AI costs $0 on the Free plan, $8 a month on Plus, $30 on Premium, $100 on Super and $200 on Ultra, with Teams starting at $30 per seat and Enterprise priced on request. Every paid plan draws on one usage pool that renews weekly, and third-party models such as Claude or GPT are billed on top at API list price.

Cursor is a coding agent and desktop code editor, now owned by SpaceX. The guide below covers what it does, what each plan includes, how to start, and how it compares with GitHub Copilot, Claude Code and OpenAI Codex.

What Cursor is and who owns it today

Cursor describes itself as “a coding agent for building ambitious software.” According to its documentation, it helps you understand a codebase, plan and build features, fix bugs and review changes, and it connects to tools such as GitHub and Slack. It runs as a desktop app, in the terminal, in Slack and in GitHub, and cloud agents can be started from a browser or phone.

Cursor is made by Anysphere. Wikipedia and other press reports say Anysphere was founded in 2022 by four MIT students: Michael Truell, Sualeh Asif, Arvid Lunnemark and Aman Sanger. Truell is generally described as CEO.

On the ownership question, CNBC reported on June 16, 2026 that SpaceX would buy Cursor in a roughly $60 billion all-stock deal. Press reports citing an SEC filing say the merger became effective on August 14, 2026, and that Cursor now operates as a subsidiary folded into SpaceX’s AI unit. Cursor has no separate stock: anyone looking for “Cursor AI stock” is looking at SpaceX shares. Our earlier piece on what the SpaceX deal means for AI coding tools covers the announcement.

Key features: Agents, models and the terminal

Cursor’s features page lists these main parts:

• Agent: plans and builds software in the desktop app, the terminal, Slack and GitHub.
• Cloud agents: agents that run in their own environments, build and test features end to end, and hand back work for review. Several can run in parallel.
• CLI: a command-line version of the agent, installed with a single curl command.
• Review: code review of pull requests in GitHub.
• Grok Bot: AI teammates that sign in to your tools and return finished work.
• Automations, Mobile and a Marketplace for rules, skills and plugins.

On models, the features page names Grok 4.7, GPT-5.6 Sol, Opus 5.5, Gemini 3.1 Pro and Cursor’s own Composer 2.5, plus an Auto option that picks for you. Which model is used matters for cost, as the next section shows. For background on an earlier model launch, see GPT-5 in Cursor.

The official pages do not describe PDF reading, image generation or mobile-app-specific tooling. The documentation only notes that some models accept images as input, so treat those uses as unconfirmed by Cursor.

Cursor AI pricing: Every plan, with limits

Cursor’s pricing documentation lists the plans below. Prices are monthly and exclude tax.

Prices checked on October 21, 2026.

A few details decide the real cost:

• One weekly pool. Cursor’s own models draw on a single allowance that covers the editor agent, the CLI, cloud agents, Grok Bot and the Grok apps. The allowance renews every week rather than once a month, so a single busy day cannot drain the whole month. The docs do not name the reset day; remaining usage and the next reset date appear in editor settings.
• Cursor models in the pool: Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5. Each model consumes the pool at a different rate.
• Plus is restricted. It includes only Cursor’s models at fixed settings (Fast off, medium effort), and it is not available for teams.
• Going over. You can wait for the reset, buy prepaid credit with top-ups or Auto Top-up, or upgrade. Cursor says requests are never downgraded in quality or speed, and a monthly additional usage limit caps overage.
• Teams and Enterprise add a Cursor Token Rate of $0.25 per million tokens on third-party model requests. Enterprise adds invoice billing, SCIM, audit logs and access controls.

How to get started with Cursor

1. Open the Cursor download page. Builds exist for macOS (Apple silicon, Intel and universal), Windows (x64 and ARM64, system or user installer) and Linux (.deb, RPM and AppImage, x64 and ARM64).
2. Install the app and sign in. The Free plan needs no payment.
3. Open a project folder and start the agent, choosing a model or leaving it on Auto.
4. Check your remaining usage and reset date in editor settings.
5. Optional: install the CLI with the curl command shown on the Cursor site, or connect Slack and GitHub.

Cursor’s download page does not list system requirements.

Limits, students and things to know

The Free plan is real but limited: Cursor does not publish a number of requests, only that agent and chat requests are capped.

Cursor’s students page says anyone can start for free and that promotions will appear at campus and online events, but it states no discount, duration or eligibility rules. Check that page before assuming a student price exists.

Self-serve plans accept major credit and debit cards. Invoices and wire transfers require Enterprise. Cancel at least 24 hours before renewal to avoid another charge. Grok API usage is billed separately and does not draw from the plan pool.

The official pages give no troubleshooting for slowness. Plus has Fast mode off by default, and higher tiers add priority access at peak times, which are the documented speed differences.

Cursor alternatives compared: Copilot, Claude Code and Codex

The three tools below are the main rivals with official pricing pages that were readable on October 21, 2026.

Prices checked on October 21, 2026.

GitHub Copilot is the closest match on price: its Pro plan costs $10 against Cursor’s $30 Premium, though the two bundle usage differently. Claude Code is a terminal-first agent bundled into Claude subscriptions; our guide to Claude Code plans, prices and installation has the detail. Codex is bundled into ChatGPT plans, so existing ChatGPT subscribers may already have access.

One structural difference: Cursor lets you pick models from several makers inside one app, but anything outside Cursor’s own models is billed at API list price on top of the plan. Copilot, Claude Code and Codex each center on their own maker’s models. Windsurf is another well-known option, but its official pricing page could not be read on October 21, so it is not compared here.

Common questions