Cursor Review 2026: Pricing, Privacy & SpaceX Ownership

Disclosure: CodeToolPilot may earn a commission if you buy through links on this page, at no extra cost to you. How we make money.

Last verified: October 12, 2026. This is an editorial research review — how we review. Prices and plans change often; confirm on the vendor’s site.

Status update: SpaceX completed its acquisition of Cursor maker Anysphere on August 14, 2026. OpenAI has announced plans to end Cursor's direct access to OpenAI models, with a proposed transition date of November 12, 2026.

Cursor is a desktop code editor built around AI agents. It is aimed at professional developers and teams who want more than autocomplete: they want to hand an agent a task, let it edit across many files and run commands, then review the result in a familiar editor. If you already use VS Code, Cursor will feel familiar from the first minute. It shares VS Code’s default shortcuts and can import your settings, keybindings and extensions in one step.

Cursor’s main appeal is that one subscription covers the whole range of AI help. You get Tab completions while you type, an in-editor Agent for multi-file changes, cloud agents that keep working after you close your laptop, and Bugbot for automated pull request review. In 2026 Cursor pushed further toward delegation with an Agents Window, Projects that coordinate many subagents, and bots that watch deployments.

This is an editorial research review based on Cursor’s pricing page, documentation, changelog and blog, plus public reporting and the vendor’s community forum. We did not run our own benchmarks. Two recent events matter here: SpaceX’s acquisition of Cursor’s maker, Anysphere, and OpenAI’s plan to end Cursor’s direct access to its models. Both are covered below.

Quick verdict

Cursor is still one of the most complete AI-first editors for developers who want agents built into everyday work, with strong team and enterprise controls. The trade-offs are cost that is hard to predict for heavy agent users and a model line-up in flux. Since the SpaceX deal, Cursor’s in-house model pool is built on Grok, and OpenAI models face a proposed cutoff. If you depend on one specific model vendor, check where things stand before you commit.

Pro: Very broad feature set in one product: Tab completions, local and cloud agents, Projects (beta), a CLI and Bugbot code review.
Pro: VS Code-style editor with one-click import of settings, keybindings and extensions.
Pro: Wide model choice plus Auto, a router that picks a model for each request in Cost, Balance or Intelligence mode.
Pro: Mature business features: SAML/OIDC SSO on Teams, plus SCIM, audit logs and repository and model access controls on Enterprise.
Con: Usage-based pools make monthly spend hard to predict. Cursor’s own docs estimate $60–$100 a month in total usage for daily Agent users.
Con: OpenAI has proposed ending Cursor’s direct access to its models, with a proposed transition date of November 12, 2026.
Con: Some forum users have raised data-use questions since the SpaceX acquisition and complained about Grok being re-selected as their default model.
Con: Extensions come from the Open VSX registry, not the VS Code Marketplace, so some extensions are missing or behave differently.

Visit Cursor

What is Cursor?

Cursor is an AI code editor from Anysphere, available for macOS, Windows and Linux. It looks and works much like VS Code, with two main views. The classic editor is for hands-on coding. The Agents Window is an agent-first workspace for running parallel agents across repositories and environments, including local, cloud and remote SSH.

Ownership changed in 2026. In April, Cursor announced a partnership with SpaceXAI, and on August 14, 2026 it confirmed that it “has officially been acquired by SpaceX.” Cursor says the deal gives it access to a very large GPU fleet for training its own models. You can see this in the product: the “Cursor Models” usage pool now contains Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5. Third-party models from Anthropic, Google, OpenAI and others are billed from a separate pool at API rates.

On August 28, 2026, OpenAI said it intends to wind down its contract supplying models to Cursor. In its help center, OpenAI says it has proposed a transition period running to November 12, 2026, and that Cursor may end access sooner. Cursor CEO Michael Truell said publicly that OpenAI models account for about 5% of Cursor traffic. OpenAI lists workarounds: your own OpenAI API key, the Codex IDE extension, or an AI gateway. It also notes that your own key works only in Cursor’s local Chat and Agent, not in Tab, Auto, cloud agents or the CLI.

Key features

Agent and the Agents Window

The Agent plans and makes multi-file changes, runs terminal commands and works with your tools. The Agents Window adds multi-workspace support, a dedicated diffs view for reviewing and committing changes, parallel agents and easy handoff between local and cloud. You can switch back to the classic editor at any time.

Tab completions

Cursor’s Tab model suggests multi-line edits as you type. Pro, Pro Plus and Ultra include unlimited Tab completions, so you can rely on everyday autocomplete without watching a meter.

Cloud agents, self-hosted machines and Projects

Cloud agents run tasks on remote machines and can be started from the editor, the web, mobile, Slack, GitHub or Linear. In September 2026 Cursor added self-hosted machines, which keep tool execution inside your own network or sandboxes. It also launched Projects in beta. In a Project, a coordinator agent plans larger bodies of work, hands tasks to subagents and keeps shared context over time.

Bugbot, Security Review and Rollouts

Bugbot reviews pull requests for bugs, security issues and code-quality problems. On September 23, 2026, Cursor added two bots for Teams and Enterprise plans. Security Review reports exploitable bugs on each pull request. Rollouts watches a change as it deploys and reports its health in each environment.

Cursor Router (Auto)

Auto routes each request to a model based on predicted complexity and task type, in Cost, Balance or Intelligence mode. Cursor reports that Auto Intelligence costs 68% less than its frontier comparison model while reaching higher user satisfaction. That figure is the vendor’s own measurement from production traffic, not an independent benchmark.

Customization and integrations

Cursor supports MCP servers, rules, skills, hooks, subagents and a plugin marketplace. A CLI is also available, and JetBrains IDEs can use Cursor’s agent through the Agent Client Protocol (ACP). An October 2026 update lets you monitor and reply to local agents from the Cursor iOS app.

Who should use Cursor

Cursor suits developers who use AI agents every day and want them inside a VS Code-like editor. It also suits teams that want one vendor for agents, code review and admin controls, and organisations that need SSO, audit logs and privacy enforcement. It is especially strong if you plan to use cloud agents and Bugbot alongside local editing.

Who should skip it

Skip Cursor if you need a fixed, predictable monthly bill. Skip it if your workflow depends on OpenAI models through a vendor-managed subscription, or if your organisation has concerns about the new ownership. If you only want light AI help, it may be more product than you need.

Strengths and limitations

Strengths

Breadth. Few editors combine autocomplete, local agents, cloud agents, PR review and deployment monitoring under one account.
Low switching cost for VS Code users. Shortcuts, themes, settings and most extensions carry over.
Model flexibility. Even after the OpenAI dispute, Cursor’s docs list models from Anthropic, Google, SpaceXAI and others, plus its own Composer model.
Enterprise readiness. Cursor’s site lists SOC 2, ISO 27001, ISO 42001 and AIUC-1 certifications. With privacy mode on, Cursor says code data is not used for training by Cursor or its model providers.

Limitations

Unpredictable spend. Included usage is measured in dollars of model usage, not requests. Frontier models drain it quickly, and Cursor’s own guidance puts power users at $200 or more a month.
Model availability risk. The proposed end of direct OpenAI access hits teams that standardised on GPT models. The bring-your-own-key fallback does not cover Tab, Auto, cloud agents or the CLI.
Ownership questions. In Cursor’s forum thread on Grok 4.5, users asked whether their privacy settings would hold after the acquisition. Cursor staff replied that Grok is subject to the same Privacy Mode controls as other models. Several users also reported Grok re-enabling itself as the default model, which staff acknowledged as a bug.
Extension gaps. Open VSX does not carry every VS Code Marketplace extension.
Growing complexity. Editor, Agents Window, Projects, cloud agents, bots and two usage pools make for a lot to learn and govern.

Pricing (as of October 2026)

Plan	Price	What you get
Hobby	Free	Limited Agent requests and access to Composer. No credit card required.
Pro	$20/month	Extended Agent limits, the Cursor Models and Other Models usage pools, unlimited Tab, cloud agents, MCPs, skills and hooks, and Bugbot on usage-based billing.
Pro Plus	$60/month	Everything in Pro with more included usage. Cursor recommends it for daily agent users.
Ultra	$200/month	Highest individual usage. Recommended by Cursor for agent power users.
Teams (Standard seat)	$40/user/month	Central billing and admin, team marketplace, Bugbot reviews, usage analytics, team-wide privacy mode and SAML/OIDC SSO.
Teams (Premium seat)	$120/user/month	Teams features with 5x the Standard seat’s Agent limits.
Enterprise	Custom	Pooled usage, invoice billing, SCIM, repository, model and MCP access controls, audit logs and priority support.

A few details matter. Usage beyond what your plan includes is billed at API rates, in arrears. On Teams and Enterprise, third-party model requests carry a “Cursor Token Rate” of $0.25 per million tokens on top of model pricing. Developers in India can choose a Start plan at ₹649 a month, which covers only the Cursor Models pool. Prices exclude tax. Prices change often, so confirm current plans and limits on Cursor’s pricing page before you buy. See Cursor pricing

Prices verified on October 12, 2026 against the official Cursor pricing page. Plans change often — confirm on the vendor’s page before you buy.

Use cases

Cursor covers everything from autocomplete to cloud agents, so the practical question is which parts a developer or team will actually rely on. These scenarios follow Cursor’s documented features as of October 2026, with the caveat that matters most for each.

Moving a VS Code workflow to an agent-first editor

A developer who lives in VS Code but wants agents to handle multi-file changes can switch with little friction, because Cursor imports settings, keybindings and extensions in one step and keeps VS Code’s default shortcuts. Tab completions cover everyday typing, while the Agent plans larger edits, runs terminal commands and leaves changes to review in a diffs view. Check your must-have extensions first, since Cursor installs them from Open VSX rather than the VS Code Marketplace.

Handing routine tasks to cloud agents

Teams that want routine work done in the background can start cloud agents from the editor, the web, mobile, Slack, GitHub or Linear, and keep them running after a laptop closes. Organizations that need tool execution to stay inside their own network or sandboxes can use the self-hosted machines Cursor added in September 2026. Cloud agents draw on usage-based pools, so agree on a spending expectation before rolling them out widely.

One vendor for agents, PR review and deployment checks

A team that wants coding agents and review from the same vendor can pair the editor with Bugbot for pull request review and, on Teams and Enterprise, the Security Review and Rollouts bots added in September 2026. Teams plans include SAML/OIDC SSO and team-wide privacy mode, and Enterprise adds SCIM, audit logs and repository and model access controls. In our assessment this suits teams comfortable with Cursor’s new SpaceX ownership; those with concerns should settle them before standardizing.

Coordinating larger projects across many agents

Developers working on multi-step efforts, such as a change that spans several repositories, can use the Agents Window to run parallel agents across local, cloud and remote SSH environments. Projects, in beta, adds a coordinator agent that plans the work, hands tasks to subagents and keeps shared context over time. Because Projects is still in beta and agent-heavy use drains included usage quickly, trial it on a bounded task first.

Using Cursor’s agent from a JetBrains IDE

Teams that work in IntelliJ IDEA, PyCharm or WebStorm can run Cursor’s agent inside the IDE through the Agent Client Protocol (ACP), where it can edit project files and run commands in the integrated terminal. That keeps JetBrains’ own refactoring and debugging tools in place while adding Cursor’s agent. Cursor’s docs say this needs a paid Cursor plan and the JetBrains AI Assistant plugin (2025.1 or later).

Alternatives

Devin Desktop (formerly Windsurf): Cognition’s VS Code-compatible IDE with an agent command center and handoff to Devin’s cloud agent.
Kiro: AWS’s agentic IDE built around spec-driven development, with credit-based pricing.
Zed: a fast, open-source native editor that runs external agents through ACP and supports your own API keys.
GitHub Copilot: an AI assistant that works inside your existing editor instead of replacing it.

Final verdict

Cursor remains a leading choice for agent-heavy development. It combines a familiar editor with cloud agents, review bots and solid enterprise controls. Budget for usage on top of the subscription, and watch two open issues: the SpaceX ownership and the OpenAI model wind-down. If neither is a dealbreaker, the free Hobby plan or the $20 Pro plan is a low-risk way to see whether its workflow fits your team.

Try Cursor

FAQ

Is Cursor free?

Yes, with limits. The Hobby plan costs nothing, needs no credit card, and includes limited Agent requests and access to Composer. Unlimited Tab completions, cloud agents and the larger usage pools start on Pro, and usage beyond a paid plan’s included amount is billed at API rates. For heavy agent users, the subscription price is therefore not always the full monthly cost.

Does Cursor use my code for training?

That depends on Privacy Mode. Cursor’s data-use page says that with Privacy Mode on, your data is not used for training by Cursor and model providers do not store or train on it, while with it off Cursor may use and store codebase data, prompts, editor actions and code snippets. Its help center says Privacy Mode is on by default for team members and admins can enforce it, but it does not state a default for individual accounts, which can turn it on in Settings. After the SpaceX acquisition, Cursor staff said on its forum that Grok models are subject to the same Privacy Mode controls as other models.

Can I still use OpenAI models in Cursor?

Only through a transition period, then in limited ways. On August 28, 2026, OpenAI said it intends to wind down its contract supplying models to Cursor, proposing a transition period that runs to November 12, 2026 and noting that Cursor may end access sooner. One workaround OpenAI lists is your own OpenAI API key, but in Cursor that works only in local Chat and Agent, not in Tab, Auto, cloud agents or the CLI. Teams that rely on GPT models should plan for the change now.

Does Cursor work with JetBrains IDEs?

Partly. JetBrains IDEs such as IntelliJ IDEA, PyCharm and WebStorm can run Cursor’s agent through the Agent Client Protocol (ACP), where it can choose models, edit files and run terminal commands. Cursor’s docs say this needs a paid Cursor plan and the AI Assistant plugin (2025.1 or later), with usage billed under your Cursor subscription. It brings in the agent, not the Cursor editor, so features such as the Agents Window remain in Cursor’s own app.

Will my VS Code extensions work in Cursor?

Most will, but not all. Cursor can import your VS Code settings, keybindings and extensions in one step, but it installs extensions from the Open VSX registry rather than Microsoft’s VS Code Marketplace. Some Marketplace extensions are missing from Open VSX or behave differently, so check the ones your workflow depends on before switching.

Cursor or GitHub Copilot?

Cursor is a separate VS Code-based editor built around agents, while GitHub Copilot works inside editors you already use, including VS Code, Visual Studio, JetBrains IDEs, Eclipse and Xcode, and connects natively to GitHub issues and pull requests. Both bill heavier AI use by consumption: Cursor charges usage beyond each plan’s included amount at API rates, and Copilot meters chat and agents in AI credits while keeping completions unlimited on paid plans. In our assessment, Cursor suits developers willing to switch editors for a more agent-centered workflow, and Copilot suits teams spread across several IDEs or built around GitHub.

Cursor Review (2026): Is the AI-First Code Editor Worth It?