What Is Cursor? Features, Pricing, Models and History: A Dated Reference for Teams

Cursor is an AI code editor and coding agent platform, now owned by SpaceX, that combines an editor, cloud agents, a command line agent and its own Composer models. Plans, models and features change every few weeks, so most of what people find about Cursor is already out of date. This page sets out what Cursor is, what each plan includes, which models it offers, how its features fit together and how the product has changed since version 1.0. It is written for developers, engineering managers and the people who approve tools for a team. Every fact comes from Cursor's own website, documentation, pricing page and changelog. Last checked: 2026-11-15.

This is an independent guide from NHI Mgmt Group, based on Cursor's public documentation as of November 2026. It is not sponsored or reviewed by Cursor. Product features change, so confirm settings against Cursor's current documentation.

Key takeaways

• Ownership changed in 2026. Cursor's blog announced on 14 August 2026 that "Cursor has officially been acquired by SpaceX", and Grok models and Grok features now appear across the plans.
• Five individual plans plus Teams and Enterprise. As of 15 November 2026 the pricing page lists Free, Plus ($8), Premium ($30), Super ($100) and Ultra ($200) per month, Teams from $30 per user per month and custom Enterprise pricing.
• Usage is a weekly pool. Each plan has an allowance that refreshes weekly, Cursor's own models draw from it, and third-party models are billed separately by usage.
• It is more than an editor. The documentation covers Agent, Plan Mode, Tab autocomplete, Cloud Agents, Bugbot, Security Review, Rollouts, Origin Git hosting, a CLI and an Agent SDK.
• Admin controls sit on Teams and Enterprise. SSO, org-wide Privacy Mode, SCIM, audit logs, model access restrictions and MDM policies are documented for organisation plans.

What is Cursor?

Cursor describes itself on its documentation home as "a coding agent for building ambitious software". In practice it is a desktop code editor with an agent built in, plus a set of services that run agents away from the developer's machine. The desktop app is the part most people know: you open a repository, ask the Agent to plan or make a change, review the diff and accept it. Around that sit Tab, the inline autocomplete; Cloud Agents, which run tasks on remote machines and can be started from a browser, phone, Slack, GitHub or Linear; Bugbot and Security Review, which comment on pull requests; and Origin, Cursor's Git hosting and repository service.

The company behind Cursor raised a $2.3 billion Series D at a $29.3 billion post-money valuation, announced on 13 November 2025, and said then that it had crossed $1 billion in annualised revenue with a team of over 300 people. On 14 August 2026 Cursor announced that it had been acquired by SpaceX. Since then Cursor's blog has announced Grok 4.6 (12 August 2026) and Grok 4.7 (21 September 2026), and the site footer links to Grok Bot terms.

Cursor's documentation is organised into Get Started, Agent, Grok Bot, Customize, Cloud Agents, Origin, Integrations, SDK and Agent SDK, CLI, and Teams and Enterprise. That structure is a fair map of the product.

Key facts at a glance

As of 15 November 2026, from Cursor's own pages.

Plans and pricing

The table below copies the plan structure on cursor.com/pricing as of 15 November 2026. Prices are in US dollars, per month, and exclude applicable taxes. The page does not show annual prices.

How usage is charged: the allowance refreshes weekly. When it runs out you can buy top-ups, and you can set spending caps. Third-party models are not included in the monthly price and are "billed separately by usage". Grok API usage is billed at standard model rates and does not draw from the plan pool. Subscriptions must be cancelled at least 24 hours before renewal to avoid another charge.

What this means for budgeting: a team that mostly uses Cursor's own models is paying for seats and pool size. A team that routes most work to third-party frontier models is paying seats plus metered usage, so spending caps matter more than the headline price.

Models available

Cursor's models page lists 23 models as of 15 November 2026. Context windows are stated for one configuration only: Grok 4.7 on Plus uses a 256k context window. For the others the page does not state a figure, so none is given here.

• Auto: Auto has three modes, Cost, Balance and Intelligence. Cost mode bills as a Cursor-model request from the weekly pool whichever model serves it. Balance and Intelligence are available on Teams and Enterprise with Cursor Router turned on; individual plans have Cost mode.
• Max Mode: extends the context window beyond the default and is billed at the model's API rate plus 20 per cent. The docs state it is available only on legacy request-based plans.
• Fast and effort settings: on Plus, Cursor's models run at fixed settings with Fast off. Fast and effort levels need Premium or higher.
• Composer: Cursor introduced Composer in version 2.0 (29 October 2025) as its first agentic coding model. Composer 2.5 is the current version listed.

Model names in Cursor follow the providers' own release cycles, so the list changes whenever a provider ships or retires a model. Check the models page before standardising on a model name in internal guidance or Rules files.

Features and modes

Two features change the trust model and are worth knowing before rollout. Remote control for local agents (6 October 2026) lets the Cursor iOS app see and reply to agents running on a computer; the changelog says it is on by default except for Enterprise organisations. Cloud agents on machines you manage (2 September 2026) move execution onto infrastructure the customer controls. Both are covered from a security angle in the Cursor Security Guide.

Supported IDEs and platforms

• Desktop app: version 3.30 on macOS (Apple Silicon ARM64, Intel x64 and Universal builds), Windows (x64 and ARM64, each with System and User installers) and Linux (x64 and ARM64 as .deb, RPM and AppImage).
• Command line: the Cursor CLI installs from the downloads page and runs in any terminal or script, including headless CI runs.
• Other IDEs: the Integrations section documents JetBrains and Xcode integrations. The CLI's ACP support is the documented route for editors that speak the Agent Client Protocol.
• Mobile and web: cloud agents can be started from a browser or phone, and the iOS app can control local agents.
• Work tools: Slack, Microsoft Teams, Jira, Linear, Notion, GitHub, GitLab, Azure DevOps and Bitbucket integrations are listed in the docs.

Enterprise and admin controls

Service accounts and API tokens used by cloud agents and the Agent SDK are non-human identities: they need an owner, a scope and a rotation plan like any other. See AI Coding Tools and Secrets Sprawl and How to audit AI coding assistant tokens and config files.

Data handling and privacy

• Privacy Mode on: Cursor states that customer data "will not be used for training by Cursor" and that it maintains zero data retention agreements with all model providers. Some models that are not under zero retention are labelled, or need an admin to opt in for a workspace.
• Privacy Mode off: Cursor may use codebase data, prompts, editor actions and code snippets to improve its AI features and train its models.
• Abuse detection exception: if prompts trigger abuse detectors, data may be stored for investigation, and providers may run risk classifiers.
• Your own API key: requests still go through Cursor's backend, so bringing a key does not bypass Cursor's servers or its data handling.
• Infrastructure: the security page states no infrastructure in China and, to Cursor's knowledge, no China-headquartered subprocessors; the subprocessor list is on trust.cursor.com.
• Assurance: AIUC-1, ISO/IEC 27001:2022, ISO/IEC 42001:2023 and SOC 2 Type II, with reports available on request, and third-party penetration testing at least annually.

The data use page links the retention policies of xAI, OpenAI, Anthropic and Meta, so review the provider policy for each model you allow, not only Cursor's.

Release history

Major versions and dated announcements from Cursor's changelog and blog. Cursor's own pages do not state a founding date or first public release date, so the history starts at version 1.0.

Limits and known constraints

• Weekly allowance: included usage refreshes weekly; heavy users hit the pool and need top-ups or a larger plan.
• Plus model scope: the pricing page states that only Grok models are available on Plus; third-party models start at Premium, billed at API list price.
• Max Mode: available only on legacy request-based plans per the models page.
• Context windows: only Grok 4.7 on Plus (256k) has a stated figure on the models page; plan around the provider's documentation for other models.
• Feature tiers: Rollouts and Security Review are Teams and Enterprise; Auto Balance and Intelligence need Teams or Enterprise with Cursor Router; SCIM, audit logs and model access restrictions are Enterprise.
• Remote control default: on by default except for Enterprise organisations, and the computer must stay on and online.
• Pricing churn: the pricing page states prices are subject to change; record the date whenever you quote a Cursor price.

Security guidance

This page describes what Cursor is and costs. Hardening is covered separately in the Cursor Security Guide, which goes through workspace trust, run modes and the sandbox, MCP allowlists, rules and hooks files, team precedence, admin roles and logging. The short version for a buyer: decide on Teams or Enterprise before rollout so Privacy Mode can be enforced org-wide, connect SSO, and treat repository files such as .cursor/mcp.json, .cursor/rules and AGENTS.md as untrusted input. The wider discipline is covered in the AI IDE and Coding Agent Security Guide and the Coding Agent Configuration File Attacks guide.

Practitioner checklist

• Record the date and plan whenever you quote a Cursor price, and recheck cursor.com/pricing before renewal.
• Choose Teams or Enterprise for any organisational use so Privacy Mode can be set to "Enforce org-wide" and SSO connected.
• Decide which model providers are allowed, review each provider's retention policy, and use Model Access Restrictions on Enterprise.
• Set spending caps before enabling third-party models, since they are billed by usage outside the plan pool.
• Decide whether remote control of local agents and self-managed cloud agents are permitted, and document the decision.
• Inventory service accounts and API tokens used by cloud agents, the CLI and the Agent SDK, with an owner and rotation date for each.
• Deploy the desktop app through MDM with allowed team IDs and extensions enforced.
• Re-read the Cursor changelog monthly and update internal guidance when defaults change.

Standards and references

• Cursor: Pricing
• Cursor: Models documentation
• Cursor: Documentation
• Cursor: Changelog
• Cursor: Changelog 2.0
• Cursor: Blog
• Cursor: Data use and Privacy Mode
• Cursor: Security

Related NHI Mgmt Group resources: Cursor Security Guide · AI IDE and Coding Agent Security Guide · AI Coding Tools and Secrets Sprawl · AI Coding Assistant Buyer's Guide