● Cheapest real entry Cursor Plus, $8/month - but Cursor models only, no third-party models and no Cloud Agents.
● Best value if you also chat Claude Pro, $17/month billed annually ($200 up front). Claude Code and the chat app draw on one pool.
● Only way to drop the 5-hour limit Claude Ultra $200 or ChatGPT Pro from $100. Nothing else removes it.
The headline prices are close. What differs is what the meter counts.
The three of these are usually compared on benchmark scores. That is not what decides your bill. What decides your bill is which other product shares your usage pool, and how long you have to wait when you hit the wall.
The plans, side by side
| Claude Code | Codex | Cursor |
|---|
| Comes with | Claude Pro, Max and Ultra | ChatGPT Free, Go, Plus, Pro, Business, Edu and Enterprise | The X Subscription (Free, Plus, Premium, Super, Ultra) |
| Entry paid tier | Pro $17/mo annual ($200 up front), or $20 monthly | Plus $20/mo | Plus $8/mo ($80/yr) |
| Mid tier | Max $100/mo, 5x Pro | Pro from $100/mo | Premium $30/mo ($300/yr), 4x Plus |
| Top tier | Ultra $200/mo, 10x Pro weekly | Pro $200 or $500/mo | Super $100 ($1,000/yr), Ultra $200 ($2,000/yr) |
| Limit windows | Five-hour session limit and a weekly limit on Pro and Max | Messages per five-hour period, and "weekly limits may also apply" | A weekly usage pool |
| Five-hour limit removed? | Yes on Ultra, for verified accounts | Yes on Pro - "Pro plans currently have no five-hour limit" | Not stated |
| Usage shared with | The Claude chat app - same limits across both | ChatGPT - Work and Codex are included in the same plan | Grok - one weekly pool across Cursor and Grok |
| Buy more without upgrading | Usage bundles | Yes on Plus and Pro | Prepaid credit for third-party models (Premium and above) |
The thing most comparisons miss: none of these meters your coding tool on its own any more. Write a long chat in the Claude app and you have less Claude Code left. Run Grok and you have less Cursor left. Budget for the pair, not the tool.
Claude Code
Verdict: the clearest limit documentation of the three, and the only one that publishes a plain annual price.
Claude Code is included with Pro, Max and Ultra. Anthropic prices Pro at $17 per month billed annually, which it bills as $200 up front, or $20 month to month. Max is $100 and described as 5x Pro. Ultra is $200, described as 10x Pro on the weekly limit.
Anthropic states that both Pro and Max carry a five-hour session limit and a weekly limit, and that these are shared across Claude and Claude Code - activity in either counts against the same allowance. There is a separate weekly limit for Fable.
Ultra is the exception. Verified Ultra accounts have no five-hour limit. Verification happens automatically based on account and payment history, and you can check it under Settings then Usage: no Current session bar means you are verified. Ultra was previously called Max 20x; existing subscribers moved across automatically at the same price.
Honest limitation: Anthropic publishes no hours-of-use estimate. You get multipliers (5x, 10x) against a Pro baseline that is itself not expressed in messages or tokens, so you cannot forecast your month before buying.
Best for: anyone already paying for Claude chat, since one subscription covers both.
Codex
Verdict: the most transparent on what a task actually costs, and the only one that publishes per-model credit rates.
Codex is included across the whole ChatGPT range - Free, Go, Plus, Pro, Business, Edu and Enterprise. ChatGPT Plus is $20 per month. Team is $25 per user per month billed annually, or $30 monthly. Pro is offered at $100, $200 or $500 per month.
OpenAI estimates local messages per five-hour period rather than giving a hard cap, and the spread by model is wide:
- GPT-6 Astra: 5 to 45 messages
- GPT-6.1 Sol: 15 to 160
- GPT-6 Sol: 15 to 150
- GPT-6.1 Terra: 35 to 285
- GPT-6 Luna: 350 to 3,000
It adds that weekly limits may also apply, without publishing the weekly figures. Pro plans currently have no five-hour limit.
Credits are published per million tokens at Standard speed - GPT-6 Luna at 2.5 input and 12.5 output, GPT-6 Astra at 250 input and 1,250 output, a hundredfold spread between the cheapest and most expensive model on the same plan. Speed costs extra against your allowance: Fast consumes 2.5x, and Ultrafast on Astra and 6.1 Sol consumes 8x.
Honest limitation: OpenAI does not publish a per-credit price - it says purchase prices depend on your plan or agreement. So the credit table tells you relative cost between models but not what a credit costs in dollars.
Best for: teams who need to model cost per task before committing, and anyone whose work sits mostly on the cheap models.
Cursor
Verdict: the cheapest way in at $8, and the one whose pricing changed most - your Cursor usage is now pooled with Grok.
Cursor’s plans are now presented as The X Subscription: Free, Plus at $8, Premium at $30, Super at $100 and Ultra at $200, with annual pricing at $80, $300, $1,000 and $2,000. Teams bills at $30, $100 or $200 per seat.
The structural change matters more than the prices. Every paid plan gets a weekly pool of usage for Cursor models - listed as Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5 - and that pool is shared across Cursor and Grok. Spend it in the chat product and it is gone from the editor.
Read the $8 tier carefully. Plus runs Cursor models only, at fixed settings (Fast off, medium effort), with no Auto, no Cloud Agents, no Automations, no Bugbot, no Origin and no third-party models. If you want Claude or GPT models inside Cursor you need Premium at $30 or above, and those are billed separately at API list price from prepaid credit on top of the subscription.
Legacy Cursor plans still run for existing subscribers only - Hobby free, Pro $20, Pro+ $60, Ultra $200, Teams Standard $40 per user and Teams Premium $120 per user - on two monthly pools rather than the new weekly one. You cannot sign up for these.
Honest limitation: on Premium and above your real bill is the subscription plus whatever the third-party models consume at list price, and Cursor does not publish an on-demand rate for the legacy plans.
Best for: developers happy on Grok and Composer, who want the lowest fixed monthly cost.
The $200 tier means three different things
All three converge on $200, and the money buys something different at each:
| $200/month buys | What you get |
|---|
| Claude Ultra | 10x Pro on the weekly limit, and no five-hour limit once the account is verified |
| ChatGPT Pro | The middle of three Pro prices ($100/$200/$500). No five-hour limit at any Pro price |
| Cursor Ultra | 50x Plus on the weekly pool, shared with Grok. The five-hour question does not arise - Cursor meters weekly |
If the five-hour wall is what breaks your day, note that ChatGPT Pro removes it at $100, half what Claude charges for the same relief.
Decision framework
- If you already pay for Claude chat → Claude Pro at $17/month annual. You are buying one pool for two tools.
- If you already pay for ChatGPT → you already have Codex. It is included on every tier down to Free.
- If you want the lowest fixed bill and Grok suits you → Cursor Plus, $8.
- If you hit the five-hour wall daily → ChatGPT Pro at $100, or Claude Ultra at $200.
- If you need to forecast cost per task → Codex, the only one publishing per-model credit rates.
- If you want Claude or GPT models inside Cursor → budget Premium ($30) plus prepaid credit, not $8.
The workflow stack that actually works
Three subscriptions is not the answer. Two, chosen so their pools do not overlap, usually is:
- Claude Pro ($17/mo annual) for the long reasoning work, with Claude Code and the chat app on one allowance.
- Cursor Plus ($8/mo) as the editor, on Cursor models only, so its pool never competes with the Claude one.
- ChatGPT Free for Codex, since Codex is included at Free and gives you a third agent at no cost when the other two are exhausted.
That is $25 a month for three coding agents on three separate meters. The usual mistake is paying $100 to one vendor and still hitting a wall at 4pm because everything drew on the same pool.
FAQ
Does Claude Code use the same limits as the Claude app?
Yes. Anthropic states the limits are shared across Claude and Claude Code, so activity in both counts toward the same five-hour and weekly limits.
Is Codex included with ChatGPT Plus?
Yes. OpenAI lists Codex as included in ChatGPT Free, Go, Plus, Pro, Business, Edu and Enterprise. Plus is $20 per month.
Which plan removes the five-hour limit?
Claude Ultra at $200 for verified accounts, and ChatGPT Pro, which starts at $100 and currently has no five-hour limit at any of its three prices.
Why does Cursor pricing mention Grok?
Cursor’s plans are presented as The X Subscription, and the weekly usage pool is shared across Cursor and Grok. Cursor models are listed as Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5.
Can I still get the old Cursor Pro at $20?
Only if you are already on it. Cursor lists Hobby, Pro, Pro+, Ultra and the two Teams plans as legacy plans for existing subscribers.
What is the cheapest way to run all three?
Cursor Plus at $8, Claude Pro at $17 a month billed annually, and Codex on ChatGPT Free - $25 a month in total.
Related
Sources
Every figure on this page was read from the vendor’s own pricing or support page on 31 October 2026. No figure here comes from a third-party roundup. Prices change often - check the linked page before you buy.