Cursor vs Claude Code: Plans, Limits and Workflow

Two $20 plans side by side: that's how this comparison used to work, and it no longer describes what's on sale. Cursor was acquired by SpaceX in August 2026 and now sells access through a shared subscription with Grok and X. Anthropic now sells an Ultra tier and applies weekly limits on top of its five-hour windows. So this guide starts with what you'd actually buy today, then gets into how the two feel to use.

The short answer

• Choose Cursor if you think in files and diffs, want autocomplete as you type, or want to switch between Grok, Composer, Claude, GPT and Gemini per task.
• Choose Claude Code if you hand off whole tasks, like working from a terminal, and want one plan that also covers Claude on the web, desktop and mobile.
• Run both if budget allows. Claude Code's extension installs inside Cursor, so this isn't an either-or decision.

Cursor vs Claude Code at a glance

Pricing and plans

Cursor pricing (as of November 2026)

Cursor's pricing page sells one subscription for Grok, Cursor and X. Free has limited Agent and Chat. Plus ($8 a month, or $80 a year) runs only Cursor's own models, with no Cloud Agents, Automations, Bugbot, third-party models or your own API keys. Premium ($30, or $300 a year) has four times Plus usage and adds those features. Super ($100, or $1,000 a year) has 15 times Plus usage and priority access at peak times. Ultra ($200, or $2,000 a year) has 50 times. Prices exclude tax.

Teams pay per seat at $30, $100 or $200 with the matching weekly usage, and can mix tiers on one team. They also pay a Cursor Token Rate of $0.25 per million tokens on eligible third-party model requests, even when using their own keys. Individual plans don't. Enterprise is custom. If you're on Hobby, Pro, Pro+ or an older Teams seat, those legacy plans keep working with their two monthly pools.

Claude Code pricing (as of November 2026)

Claude Code comes with a paid Claude plan; Anthropic's pricing page lists it on all of them. Pro is $20 a month, or $17 a month with $200 billed up front. Max is $100 a month with five times Pro usage. Ultra is $200 a month with 10 times Pro weekly usage and no five-hour limit for verified accounts. Max and Ultra also include Fable, Anthropic's top model, for up to half the weekly limit; on Pro it's paid with usage credits.

Team Standard seats are $20 per seat a month billed annually or $25 billed monthly, and Premium seats are $100 or $125 with five times the usage. Enterprise is $20 per seat a month plus usage at API rates, billed annually. The alternative is a Claude Console account with pay-per-token API billing. Anthropic's cost guide puts the average across enterprise deployments at about $13 per developer per active day, and says it varies widely with model choice and codebase size. Treat that as a planning number, not a quote.

Usage limits and what happens when you hit them

Neither vendor gives you a message count. Cursor's models draw from the pool at different rates, and Claude usage depends on the model, how long your chats run and how large your files are.

Cursor gives each paid plan one pool that refreshes weekly. Agent in the editor, the CLI, Cloud Agents, Grok Bot and the Grok apps all draw from it. When it runs out you wait, upgrade, or spend prepaid credit at per-model rates, and Cursor says requests are never downgraded in quality or speed. You can set a monthly cap on additional usage.

Claude plans use a rolling five-hour window plus a weekly limit, and your chats in Claude count against the same pool as your terminal sessions. A long research chat at lunch can shorten your afternoon of coding. Hit the limit and you can wait, upgrade, turn on usage credits at standard API rates, or switch to a Console account. One trap: the monthly API credits on Max and Ultra can't be spent in Claude Code.

Models

This is the sharpest difference, and the one older posts miss. Cursor's included pool covers only Grok 4.7, 4.6, 4.5 and Composer 2.5. Everything else, including Claude, GPT and Gemini, is paid from prepaid credit at API list price, or through your own key on Premium and above. Cursor lists Claude Sonnet 5.5 at $2 per million input tokens and $10 per million output, the same list rate as Anthropic's API page. So you get a wide menu, and the Claude part of it isn't subsidized by your subscription.

Claude Code goes the other way. It runs Claude models only, across Haiku, Sonnet, Opus and Fable, reachable through your Claude plan, the API, Amazon Bedrock, Google Cloud or Microsoft Foundry. You lose the menu. In return, a $20 plan covers Claude usage up to its limits. If your work is Claude-heavy, that bundling is often the whole decision.

Day-to-day workflow

Cursor is where you sit. It's based on the VS Code codebase and imports your settings in one click, and Tab suggests edits as you type, including across files. It has Agent, Plan and Ask modes, and Plan Mode writes a plan you can edit before any code changes. Checkpoints snapshot your files so you can restore after a wrong turn. There's also a CLI, and JetBrains users can reach Cursor's agent through the Agent Client Protocol.

Claude Code is something you talk to. You describe a task, it reads the code, edits files and runs commands. The terminal is the fullest version; there are also VS Code and JetBrains extensions, a desktop app, the web at claude.ai/code and a mobile app. It has plan mode and checkpoints with rewind too, though a file that only bash commands changed isn't tracked, so keep using Git. There's no Tab-style code autocomplete. If you still write a lot of code by hand, you'll feel that.

Configuration has converged. Both support skills, subagents, hooks, MCP servers and plugins, and both can read AGENTS.md. Cursor even loads skills from .claude/skills, so a team using both doesn't have to write everything twice. Whichever you pick, good habits carry over: see our best practices for AI coding and, for Cursor specifically, our Cursor prompts collection.

Permissions and safety

Both now offer a classifier as a middle ground between approving everything and approving nothing. Cursor's Auto-review runs allowlisted calls, sandboxes shell commands where it can and sends the rest to a classifier. Its docs say agents can edit workspace files without approval (config files excepted) and that changes save to disk immediately, so keep everything in version control.

Claude Code has Manual mode, which asks before most edits, shell commands and network access, plus acceptEdits, plan, auto and bypassPermissions. From v2.1.283, auto mode is the starting mode for interactive terminal and VS Code sessions. If you want to approve every edit, switch to Manual on purpose.

On data, Cursor says it won't train on your data with Privacy Mode on, and that it may if you turn it off. Anthropic gives Free, Pro, Max and Ultra accounts a setting for whether data can train models, while Team, Enterprise and API use isn't trained on unless a customer opts in. Check both settings before pointing either at client code.

Cloud agents, review and automation

Cursor Cloud Agents run in isolated VMs, can build and test, and start from the desktop app, web, iOS or Slack. They work with GitHub, GitLab, Bitbucket and Azure DevOps, and they need Premium or above. Automations trigger them on a schedule or from GitHub, GitLab, Slack, Linear and webhooks. Bugbot reviews pull requests, and Origin, Cursor's own git host, is in early beta.

Claude Code cloud sessions run on Anthropic-managed VMs at no separate VM charge; they count toward your plan limits and need a Claude subscription, not an API key. Anthropic-hosted cloning and pull requests require GitHub, so a GitLab shop will find this tighter. For review, /code-review runs on any plan, while the managed Code Review service is Team and Enterprise only and is billed on token usage through usage credits, on top of the plan. Scheduled tasks, GitHub Actions and GitLab CI/CD cover automation.

Who each is for

• Cursor fits developers who review diffs in an editor and still type a lot, teams that want several model vendors behind one bill, and shops on GitLab, Bitbucket or Azure DevOps that want cloud agents.
• Claude Code fits developers who delegate multi-file tasks, terminal users, people who already pay for Claude, and teams that want the same agent on laptops and in CI.

Limits of each

• Cursor: Claude and other third-party models cost extra on every plan, and Plus doesn't offer them at all.
• Cursor: your weekly pool is shared with the Grok apps, and the plans bundle X and Grok perks you may not want.
• Cursor: Teams and Enterprise pay the Cursor Token Rate on eligible third-party models, even with their own keys.
• Claude Code: Claude models only, and no free tier.
• Claude Code: the five-hour window can stop you mid-afternoon on Pro and Max, and chat usage eats into coding usage.
• Claude Code: cloud sessions and mobile need a Claude subscription, and hosted cloud cloning needs GitHub.

Which should you choose

Ask how you spent yesterday. If you were in the editor, accepting and adjusting suggestions line by line, Cursor will feel right, and Premium at $30 is the realistic starting point because Plus drops cloud agents and third-party models. If you were writing a paragraph of intent and checking the result ten minutes later, start Claude Code on Pro at $20 and move to Max only if you hit limits often.

Then run the same real task through both: one bug fix touching three files, in your own repository, for a week. Watch which limit you hit first and how much you had to correct. Your repository will tell you more than any table, including this one. For how far to trust either agent once it's running, our prompting techniques for code quality are a good next read.

Alternatives worth a look

If neither fits, our list of AI pair programming tools covers GitHub Copilot, OpenAI Codex, Kiro and others, with current plans. Prefer something open source? See open-source AI coding tools.

Frequently Asked Questions

Is Cursor or Claude Code cheaper?
It depends on the models you want. Cursor Plus is $8 a month but only runs Cursor's own models (Grok and Composer). Using Claude models in Cursor needs Premium at $30 a month plus prepaid credit at API list price, or your own API key. Claude Pro is $20 a month, or $17 a month billed annually, and its included usage covers Claude Code with Claude models. Prices as of November 2026.

Can I use Claude models in Cursor?
Yes, on Premium and above. Third-party models such as Claude, GPT and Gemini are not part of Cursor's included weekly pool. You pay for them from prepaid credit at the model's API list price from the first request, or you add your own Anthropic API key and Anthropic bills you directly. The Free and Plus plans do not include third-party models.

Can I use Claude Code inside Cursor?
Yes. Claude Code's editor extension installs in VS Code and in forks such as Cursor, and a Pro, Max or Ultra plan covers it. Usage counts toward the same limits as the rest of your Claude plan, so you pay for the Claude plan separately from any Cursor plan.

Is there a free version of Claude Code?
No. Anthropic's pricing page lists Claude Code on Pro, Max, Ultra, Team and Enterprise plans, not on Free. You can also pay per token through a Claude Console account. Cursor does have a Free plan, with limited Agent and Chat usage.

Does Cursor still have a $20 Pro plan?
Only for existing customers. Cursor is now sold as part of The X Subscription, with Free, Plus, Premium, Super, Ultra, Teams and Enterprise plans. Its docs describe Hobby, Start, Pro, Pro+, the old Ultra and the older Teams seats as legacy plans that keep working for people already on them.

Who owns Cursor and Claude Code?
Cursor says it was acquired by SpaceX, in a company post dated August 14, 2026, and its site footer still carries the Anysphere name. Claude Code is made by Anthropic.