Home › Rules › Cursor vs Claude Code vs Codex

Cursor vs Claude Code vs Codex: which AI coding tool should a beginner use?

Prices and plans from Cursor's, Anthropic's and OpenAI's own pricing pages, checked Oct 27, 2026. Sources below.

Comparison
Cursor
Claude Code
Codex
Beginners
Pricing

Short answer: if you've never coded and want an app without installing anything, start with a browser builder like Lovable, Bolt, Base44 or Replit (see below). If you want the code on your own computer, all three tools here work. Cursor is a code editor app with an AI agent built in, and it has a free plan. Claude Code comes with a Claude Pro plan ($20 a month). Codex comes with ChatGPT, and OpenAI's pricing page lists Codex on the Free and Go plans too (in the desktop app, "subject to rollout"). Already paying for Claude or ChatGPT? Try the tool that comes with that plan before you pay for another one. All three read an AGENTS.md rules file, so you can switch later without starting over.

Side-by-side breakdown

	Cursor	Claude Code	OpenAI Codex
What it is	A code editor app with an AI agent built in. Cursor calls itself "the AI editor and coding agent."	Anthropic's coding agent. It reads your project, edits files and runs commands.	OpenAI's coding agent, part of ChatGPT.
Where it runs	Desktop app for Mac, Windows and Linux. Also a command-line tool, and cloud agents on Premium and up.	Terminal, VS Code and Cursor extension, JetBrains plugin, a desktop app, and the web at claude.ai/code.	The ChatGPT desktop app (Mac, Windows, Linux), the Codex CLI, an IDE extension, the web and iOS.
Who it's for	People who want to see the code and click through the changes in an editor	People who already pay for Claude, or are OK typing in a terminal (the desktop app skips that)	People who already pay for ChatGPT
Free tier	Yes, $0: "limited agent requests"	No. Claude's Free plan doesn't include Claude Code.	Yes, $0: listed for the desktop app, "subject to rollout"
Cheapest paid plan (checked Oct 27, 2026)	Plus, $8/month (Grok models only). Premium is $30/month.	Pro, $20/month, or $17/month if you pay $200 for a year	Go, $8/month (desktop app, light tasks). Plus is $20/month and adds the CLI, IDE extension, web and iOS.
Bigger plans	Super $100, Ultra $200 a month	Max, from $100 a month (5x or 20x Pro's usage)	Pro at $100, $200 or $500 a month
Rules file it reads	`.cursor/rules/*.mdc` or `AGENTS.md`	`CLAUDE.md` (or `AGENTS.md` if there's no CLAUDE.md)	`AGENTS.md`

Prices are per month in US dollars, before tax, from each company's own pricing page. All three plans have usage limits; see usage limits.

Pick this if...

Pick Cursor if

You want a normal app you download and open, with your files on the left and the AI chat next to them.
You want to start for $0. The Free plan has "limited agent requests", so it's good for trying Cursor out on a small project.
You want one place to use more than one AI company's models. Watch the plan, though: Plus ($8) is Grok models only. On Premium ($30) and up, other companies' models are billed separately at API list price, on top of the plan. Cursor's own pricing FAQ recommends Premium for new AI users.

Pick Claude Code if

You already pay for Claude Pro or Max. Claude Code is included, so it costs you nothing extra to try.
You're OK with a terminal, or you'd rather use the Claude desktop app's Code tab. Anthropic's docs say the desktop app needs a paid plan.
You like Cursor's editor too. Claude Code has an extension for VS Code and Cursor, so you can run it inside Cursor.

Pick Codex if

You already pay for ChatGPT Plus or Pro. Codex is part of it, in the desktop app, the CLI, the IDE extension, the web and iOS.
You want to try an agent for $0. OpenAI lists Codex in the desktop app on the Free and Go ($8) plans, "subject to rollout", so it may not show up for everyone yet.
You're on a cheap plan and only need small jobs. OpenAI describes Go as "lightweight coding tasks" and Plus as "a few focused coding sessions each week."

Not sure yet? Cursor and Codex both have $0 options, so you can try both on the same small project before paying. If you already have a Claude or ChatGPT plan, try the tool that comes with it first. If you're coming from a builder like Lovable, our guide to moving a Lovable project to Cursor or Claude Code covers the switch.

Usage limits, in plain words

Cursor: every plan has a pool of AI usage shared between Cursor and Grok. It refreshes weekly. Higher plans get a bigger pool (Premium is 4x Plus, Super 15x, Ultra 50x). If you run out, you can buy top-offs or wait for the refresh.
Claude Code: it uses your Claude plan's usage. Anthropic's pricing page says "usage limits apply" and that Max gives you 5x or 20x more usage than Pro.
Codex: OpenAI says how far your plan goes depends on the model, how big the task is, and whether it runs on your computer or in the cloud. On Plus, limits are counted per five-hour window. OpenAI says Pro plans currently have no five-hour limit. You can also use Codex with an API key and pay per use at API prices, but cloud features like GitHub code review aren't available that way.

None of them give a simple "X requests a month" number for these plans. Check the usage page in each tool's settings to see what you have left.

How each tool reads your rules file

A rules file is a plain text file in your project that tells the AI how you want things done, so you don't repeat yourself every chat. Here's what each tool reads, from its own docs:

Tool	Reads	Good to know
Cursor	`.cursor/rules/` (`.mdc` files) and `AGENTS.md`	Rules in `.cursor/rules` must end in `.mdc`; a plain `.md` file there is ignored. `AGENTS.md` works in the project root and in subfolders. Our Cursor rules examples have ready-made .mdc files.
Claude Code	`CLAUDE.md`, `.claude/CLAUDE.md`, `~/.claude/CLAUDE.md`	If your project has an `AGENTS.md` and no `CLAUDE.md`, Claude Code reads the `AGENTS.md` (v2.1.277 or later). If you have both, by default it reads only `CLAUDE.md`, unless you import `AGENTS.md` (below) or change the Project instructions setting. Run `/init` to have it write a starter CLAUDE.md. See our CLAUDE.md examples.
Codex	`AGENTS.md`, plus `~/.codex/AGENTS.md` for all your projects	It reads files from the project root down to the folder you're in, and the closer file wins. It stops adding files at 32 KiB by default. See our AGENTS.md examples.

One rules file for all three

Put your rules in AGENTS.md in the project root. Cursor and Codex read it as is. Here's a short starter you can copy and fill in:

AGENTS.md (project root)

# AGENTS.md

## About this project
- A small web app for [who it's for] that lets them [the main thing it does].
- Built with [your stack, for example Next.js + Supabase].

## How to work
- Explain what you plan to change before you change it.
- Make one small change at a time, then tell me how to test it.
- Don't add new packages without asking me first.
- Never put API keys or passwords in the code. Use environment variables.

## Commands
- Install: npm install
- Run locally: npm run dev
- Tests: npm test

If you also want a CLAUDE.md for Claude-only notes, start it with @AGENTS.md. Anthropic's docs say Claude then reads the imported file first, then the rest:

CLAUDE.md (next to AGENTS.md)

@AGENTS.md

## Claude Code
- Use plan mode for any change that touches more than 3 files.

More on which file wins and when: AGENTS.md vs CLAUDE.md and Cursor rules vs AGENTS.md. Or build a rules file for your project in the free rules generator.

Matching kit: MVP Spec-to-Build Pack ($7)

The free rules set the boundaries. The MVP pack gives you the build order: a 1-page spec to fill in first, then 7 prompts in sequence (skeleton, review, login, database, security, one feature at a time, polish), plus a "did I leak user data?" checklist.

See the MVP Spec-to-Build Pack ($7) →

Instant download. 7-day refund, no questions asked.

Don't want to code at all? A builder might fit better

Cursor, Claude Code and Codex work on code files on your computer (or a copy in the cloud). Browser builders are different. You chat in a web page, and they build and host the app for you. People search for "Lovable vs Claude Code" and "Base44 vs Replit" because both kinds get called vibe coding. Here's how the builders' own pricing pages read today:

Builder	What it says it is	Free tier	Cheapest paid plan (checked Oct 27, 2026)
Lovable	Chat to build websites and web apps, "no technical knowledge needed"	Yes: 5 build credits a day (up to 30 a month)	Not shown on the page we could fetch; check their pricing page
Bolt	AI website and app builder	Yes: 1M tokens a month (300K a day)	Pro, $25/month billed monthly (cheaper billed yearly)
Base44	No-code AI platform: front end, back end, database and login built for you	Yes: 25 message credits a month, up to 5 apps	Starter, $16/month billed yearly (100 message credits a month)
Replit	Describe what you want, then create, launch and grow apps and sites in one place	Not listed on the pricing page; check their pricing page	Core, $20/month, or $18/month billed yearly

We've already compared the builders head to head with real prices: Lovable vs Bolt vs Replit and Base44 vs Lovable. Each one also has a free rules template: Lovable, Bolt, Base44 and Replit.

If you start in a builder and later want more control over the code, you can move the project to Cursor, Claude Code or Codex. Our Lovable to Cursor or Claude Code guide walks through it.

Which builder should you use?

Base44 (try Base44): an all-in-one builder for internal tools, portals and simple business apps, with the least setup. Less control over the code than Lovable.
Bolt (try Bolt): the pick for a phone app (iPhone and Android via Expo) or more hands-on control of the code. Token-based, so bigger projects cost more per message.
Lovable (lovable.dev): our default for a first web app with login, a database and payments. Credits per message, and fix-retry loops burn them.

All three have free plans, so try your pick on a small project first. More detail: Lovable vs Bolt vs Replit and Base44 vs Lovable.

Some links are affiliate links; we may earn a commission at no cost to you. The Base44 and Bolt links are affiliate links; the Lovable link is a plain link. It never changes our pick.

Next step for Cursor, Claude Code or Codex: Agent Skills

Skills are folders your AI coding agent loads only when the job comes up, and Claude Code, Cursor and OpenAI Codex all read the same SKILL.md format. The Vibe Coder Skills Pack ($9) is 6 ready-to-copy SKILL.md skills (fix-one-bug, stop-the-loop, leak-check, ship-check, explain-project, builder-handoff) for Claude Code, Cursor and OpenAI Codex. Also included in the 6-kit bundle.

Get the Skills Pack for $9 on Gumroad → See what's inside

One .zip of skill folders plus one markdown file. 7-day refund, no questions asked.

Best value

All 6 kits for $29 — or $20.30 with code WELCOME30 (through Nov 30)

Whichever tool you pick, it works better when it knows the plan. The kits tell your AI what to build, step by step: a small game, a Discord bot, an email-triage agent, an MVP, or a cleanup of messy code.

Get all 6 kits for $29 → See what's in each kit

$29 instead of $48 for the six kits bought separately; WELCOME30 takes 30% off that at checkout ($20.30). Instant download from Gumroad. 7-day refund, no questions asked.

AGENTS.md vs CLAUDE.md (which file each tool reads)
Cursor rules vs AGENTS.md
Lovable vs Bolt vs Replit for beginners (real prices)
Base44 vs Lovable for beginners
Move a Lovable project to Cursor or Claude Code
Plan mode guide (6 tools + 6 plan prompts)
Best AI tool to fix and refactor existing code
Best AI tool to build a web app MVP

Sources

All prices and plan details checked Oct 27, 2026 on each company's own pricing and docs pages. Prices are in US dollars, before tax, and change often. If something here no longer matches, tell us at [email protected].