Cursor review

What is Cursor?

Cursor is an AI coding environment for people who want to work directly with a project’s source code. It combines local editing, terminal work, and remote agents, with rules, extensions, and source-control integrations for adapting the workflow. That makes it especially relevant to an existing application or a new project whose builder expects to maintain the code personally. It is less of a guided publishing service than a place to develop and review software. Beginners can learn with it, but understanding files, dependencies, and changes is part of the job rather than something the interface completely removes.

What’s changed in 2026

SpaceX acquired Cursor, which continues operating under the Cursor name. The X Subscription introduced new bundled plans with a shared weekly usage pool, while existing subscribers can keep legacy plans unless they opt to switch. The distinction between included Cursor-owned models and separately billed third-party models is central to the new offer. OpenAI has also announced a future contract wind-down, creating uncertainty for workflows tied to that model family; access has not ended as part of that announcement. The old referral program has ended. An older review centered on the former Pro subscription therefore no longer describes the current choice for a new customer.

What it’s like to use

The most useful aspect is that the assistant stays close to code you can read and change. Developers can review a plan, ask for a bounded implementation, and inspect the resulting differences before continuing. Many builders prefer that involvement to accepting a large opaque build. Model selection can help when an answer is weak, although availability and cost depend on the plan. We would use the assistant to explain unfamiliar logic, prepare a proposed change, and implement it in stages. The practical benefit comes from shortening work you can still evaluate, rather than assuming that an agent’s confidence proves a change is correct.

Where it falls short

Sustained agent work can exhaust the included allowance sooner than expected. Output quality and responsiveness also vary: a model that handles ordinary edits well may struggle with a difficult correction, and automatic selection is not always the best fit. New subscription restrictions can weaken the flexibility that attracted some builders to the editor. The entry paid tier excludes third-party models, while higher-tier use can add charges from the first request. These financial boundaries sit alongside a learning curve around code review, tools, and deployment. Someone who wants a complete hosted app without engaging with its implementation may find this environment unnecessarily demanding.

Pricing and credits in practice

The current new-customer plans are Plus at $8 per month, Premium at $30, Super at $100, and Ultra at $200. Teams starts at $30 per user per month rather than a fixed total for the organization. A limited Free plan is also available. Those prices do not mean all model usage is included: Plus is restricted to Cursor-owned models, and eligible third-party models on higher tiers are separately metered. Compare the weekly pool with the way you actually work. Intensive sessions can feel different from evenly spread usage, and retaining a legacy plan may be preferable to switching without checking its model and quota rules.

Who should use it (and who shouldn’t)

We would choose Cursor for a developer or committed learner who wants to own the implementation and review changes as they happen. Existing repositories, ongoing feature work, and small staged personal projects are natural fits. It is less attractive for someone who mainly wants bundled access to third-party models or uninterrupted premium-model work at a fixed subscription cost. Devin Desktop, formerly Windsurf, offers another agent-oriented coding environment. Dyad provides a local app-building alternative, while Lovable is easier to consider when the priority is a guided first web app rather than managing an editor and development process.

Bottom line

Cursor remains a useful option when AI assistance supports a development process you control. Its value is not simply that it can generate a large amount of code, but that you can inspect, redirect, and maintain the changes in a familiar workspace. We would assess the current plans using the exact models and working rhythm the project needs. The acquisition has not ended the product, but pricing and model availability have changed enough to make old assumptions unreliable. Choose it for the editor and review workflow, with a clear understanding of the included pool and the extra charges attached to your preferred models.

Developer Resource

If you use Cursor to build AI-native features, a unified OpenAI-compatible API layer can make model switching, API key management, and inference cost planning simpler.

Explore ValueAPI