Cursor

What Cursor actually is

Cursor isn’t a chatbot you keep open next to your code editor — it is the code editor, built as a fork of VS Code with AI wired directly into how you write. It’s not a general-purpose assistant like ChatGPT or Claude with a coding mode bolted on; everything about it is built around one job, writing and editing real codebases.

There are three ways you actually interact with it day to day. Tab is inline autocomplete — it predicts your next edit as you type, often across multiple lines or files at once, not just the next word. Cmd+K gives you a targeted, one-off edit: highlight a block, describe the change, and it rewrites just that piece. Agent is the autonomous mode — describe a task in plain language and it searches your codebase semantically, edits whichever files the task actually touches, runs terminal commands to test its own work, and keeps going without a cap on how many steps that takes. Under the hood it isn’t locked to one model either — it routes across models from Anthropic, OpenAI, and others depending on the task, plus some models Cursor has built itself.

Pricing sits on top of that core product as six tiers: Hobby (free), Pro ($20/month), Pro+ ($60), Ultra ($200), Teams ($40 standard or $120 premium per seat), and Enterprise (custom). Every paid plan below Teams gives you the same editor and the same three modes — the difference is how much usage-based credit you get each month, not which features you can see. Pro’s $20 credit pool covers most individual daily use; Pro+ and Ultra exist specifically for people who burn through that regularly, at 3x and 20x the usage respectively.

Where it’s genuinely good

The agent is the actual product here, not a chat window bolted onto an editor. Point it at a task and it searches your codebase semantically — by meaning, not just keyword matching — makes the edit across however many files that requires, and runs a terminal command to check its own work, without a hard limit on how many tool calls that takes. Tab, the autocomplete layer, runs on the technology Cursor got when it acquired Supermaven, and it’s unlimited on every paid plan, which is the single feature multiple pricing guides point to as justifying the $20 on its own.

The MCP ecosystem is the other real strength. Because Cursor adopted Anthropic’s Model Context Protocol early, there’s now an actual marketplace of third-party connectors — cloud infrastructure, CRM data, Google Workspace, security scanning — that plug into the agent directly. Combined with repo-level rules files that encode your team’s actual conventions, the result feels less like an autocomplete tool and more like an agent that’s actually been onboarded to your codebase.

Where it falls short

The credit-based billing is the thing to understand before you rely on this daily. Cursor switched from a fixed request count to usage-metered credits in mid-2025, and the backlash was real enough that the company rolled back part of the change and refunded affected users — a heavy “Max Mode” session across a large repo can burn through a chunk of your monthly pool in one sitting, which makes cost harder to predict than a flat subscription. Reliability took a separate hit in April 2025, when an AI-powered support agent invented a cancellation policy that didn’t exist, prompting real cancellations before Cursor apologized and issued refunds — worth knowing about even though it’s a support-desk incident rather than a coding-agent one.

The SpaceX acquisition, announced in June 2026, is the newest wrinkle. Nothing about the product has visibly changed yet, but it’s the kind of ownership change that’s worth watching if you’re building a long-term workflow around this tool specifically.

Which plan to actually get

Start on Hobby to get a feel for the agent and Tab — it comes with a one-week Pro trial the first time you sign up, so you’ll see the full experience before deciding. Move to Pro once you’re using it daily; unlimited Tab plus a $20 credit pool covers the large majority of individual developers without hitting a wall. Pro+ and Ultra only make sense if you’re regularly running heavy multi-file agent tasks that chew through Pro’s credit pool — for occasional heavy days, staying on Pro and eating the odd slow request is cheaper than jumping tiers. Teams is worth it once you need centralized billing, SSO, and a privacy mode that guarantees your code isn’t used for training — those controls, not a better editor, are the actual reason to pick it over individual Pro seats.

Ease of use

If you’re already in VS Code, switching over will be quite straight forward— Cursor imports your extensions, keybindings, and themes on install, so the editor itself isn’t what you’re learning. What takes actual ramp-up time is the agent workflow around it: deciding when to use targeted inline edits versus letting the agent run autonomously, setting up rules files, and understanding how the credit system actually spends your monthly allowance. None of that is complicated once you’ve done it once, but it’s a real curve, not an instant one.

Who it’s actually for

Developers, full stop — this isn’t a general-purpose assistant with a coding mode, it’s a coding tool that happens to also chat. It’s the strongest pick if you want the editor and the agent built as one product rather than an extension bolted onto an existing setup, and if you’re comfortable with usage-based billing instead of a flat cap. If your team is already standardized on a specific model provider’s CLI tooling, or you’d rather keep your existing editor and add AI as a plugin, that’s a real reason to look at the alternatives instead.