Compare

Cursor Ultra vs Claude Code Max (2026): Which $200 Plan Wins?

Cursor Ultra and Claude Ultra both cost $200 a month. Usage, models, IDE vs terminal, and which heavy coder gets more per dollar, from vendor pages.

By Engineer Foodie Desk Oct 26, 2026 ✓ Sources checked Oct 26, 2026

9 min read

Cursor plans and model prices checked Oct 26, 2026 in Cursor’s models and pricing docs. Claude plan features checked Oct 26, 2026 on claude.com/pricing. Other dates are under Sources. No affiliate links (how we work).

Cursor Ultra and Claude Ultra (formerly Max 20x) both cost $200 a month, but they buy completely different things. Ultra buys a big weekly pool of Cursor’s own models (Grok 4.7, 4.6, 4.5 and Composer 2.5). Claude Ultra buys Claude: 20 times Pro’s per-session usage, in Claude Code and the Claude apps. And here’s the part that settles it for most people: Claude isn’t in Ultra’s pool at all. On Cursor, every Claude request is billed at Anthropic’s API list price from prepaid credit, on top of the $200.

So if “Claude Code Max” is in your search because you want Claude all day, Claude Ultra is the plan. If you want an editor full of agents running Cursor’s models, Ultra is. If you want both, you’re paying for both.

Quick answer (Oct 26, 2026)

You want Claude models all day, in a terminal or IDE: Claude Ultra. Claude Code is included and shares one pool with Claude chat.
You want an AI editor with Cloud Agents, Automations and Bugbot, and you’re fine with Grok and Composer: Cursor Ultra.
You want Claude inside Cursor: you can run Claude Code in Cursor on a Max plan (it counts toward your Max limits), or pay API list price per token in Cursor’s own agent.
You hate surprise bills: both have a hard included allowance; neither publishes its exact size.

The $200 table

Plan	Price	Usage	Models	Surface
Cursor Ultra	$200/mo or $2,000/yr	50x Plus, weekly	Grok, Composer	Editor, CLI, cloud
Claude Ultra	$200/mo	10x Pro weekly, no 5h limit	Claude	Terminal, IDE, apps

USD. Cursor: models and pricing docs, “Prices exclude tax” (verified Oct 26, 2026). Claude: Max plan article, monthly only, web prices, “Mobile pricing may vary” (verified Oct 24, 2026); features from claude.com/pricing (verified Oct 26, 2026).

Cursor sells Ultra yearly ($2,000, or $400 less than 12 monthly payments); Anthropic sells Max monthly only.

What $200 buys on Cursor Ultra

Cursor’s docs describe Ultra as “Everything in Super, with 50x Plus usage and earliest access to new products and models”, and say it “suits power users running several agents or Automations” (verified Oct 26, 2026). In practice:

One weekly pool for Cursor models: Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5. It’s “shared across everything on your subscription: Agent in the editor, CLI, Cloud Agents, Grok Bot, and the Grok apps”, and “It refreshes weekly”. Models draw from it “at different rates”.
The size is relative. Ultra is “50x Plus”. What Plus itself holds isn’t published, so 50x of an unknown is still unknown. Your remaining usage and next reset show in editor settings and under Usage at grok.com/dashboard.
Everything from Premium and Super: Cloud Agents, Automations, code hosting on Origin, Bugbot code review, third-party models at API list price, and Super’s “priority model access at peak times”.
Own API keys work on Ultra, but “Custom API keys only work with chat models. Tab completion continues using Cursor’s built-in models”, and your provider bills you directly (Cursor API keys help, verified Oct 23, 2026).

Then the Claude part. “Third-party models like Claude, GPT, and Gemini aren’t part of the included pool. On Premium and above, they’re available in Cursor at the model’s API list price from the first request, paid from your prepaid credit.” Listed rates per million tokens (input / output), verified Oct 26, 2026:

Model	Input	Output
Claude Opus 5.5	$4	$20
Claude Sonnet 5.5	$2	$10
Claude Fable 5.1	$10	$50
Claude Haiku 5.5	$0.10	$0.50

Source: Cursor models and pricing (verified Oct 26, 2026). Older Claude models are listed but “Hidden by default”. Individual plans don’t pay Cursor’s $0.25-per-million Token Rate; Teams and Enterprise do.

Still on the old Cursor Ultra? Legacy Ultra ($200) has two monthly pools, one of them for third-party models “drawn down at each model’s API price”, plus on-demand usage. New customers can’t buy it, and “If you switch… the change is permanent” (Cursor pricing help, verified Oct 24, 2026).

Cursor’s own docs say each new paid plan “drops the included third-party model allowance” the legacy plans had. So the $200 top tier of a coding editor, as sold on Nov 4, 2026, includes zero Claude. That’s a sad thing to write about a product that a lot of people picked up precisely to use Claude in a good editor.

Source: Cursor pricing help, verified Oct 24, 2026

Watching Cursor get folded into The X Subscription, shared pool with the Grok apps and all, feels like the Michael Scott Paper Company getting bought back by Dunder Mifflin. The product’s still there. The name on the check changed, and so did what’s in the box.

What $200 buys on Claude Ultra

Anthropic’s version is simpler to describe and just as vague on absolute numbers:

10x Pro weekly, no 5h limit. Claude Ultra (formerly Max 20x) gives “10x Pro usage every week” and verified accounts get “No 5-hour limit” (claude.com/pricing, verified Nov 1, 2026). The weekly limit has a fixed reset time assigned to your account (Max plan article, verified Oct 24, 2026).
Shared with claude.ai. “These limits are shared across Claude and Claude Code, meaning all activity in both tools counts against the same limits.” IDE use (VS Code, Cursor, JetBrains) counts too (Claude Code with Pro or Max, verified Oct 23, 2026).
Every Claude model, with one asterisk. Opus, Sonnet and Haiku are included; Fable on Max is “50% of weekly limits” per the pricing grid, and Max has “a separate weekly limit for Fable” (claude.com/pricing, verified Oct 26, 2026; support article verified Oct 23, 2026).
Priority access at high traffic times, and context “Up to 1M varies by model” (pricing grid, verified Oct 26, 2026).
$200 a month in API credits “to build on the Claude Platform”. Those are for apps and scripts you build; they “can’t be used in Claude Code or the Claude apps” (pricing grid verified Oct 26, 2026; support article verified Oct 23, 2026).

No message count is published: Anthropic’s FAQ says “there’s no fixed message count” (claude.com/pricing, verified Oct 26, 2026).

Claude Code lives in a terminal: the beat-up RV in the desert from Breaking Bad, next to Cursor’s shiny lab. Both cook.

Overage: what happens at the cap

Cursor Ultra. When the weekly pool runs out, you can “Wait for the reset”, “Use prepaid credit” at the same per-model rates, or “Upgrade your plan” (Ultra is already the top individual tier, so really it’s the first two). “Requests are never downgraded in quality or speed.” A monthly additional usage limit caps paid extra usage, and prepaid credit “stays available until used or until it expires (currently 365 days after purchase)” (docs verified Oct 26, 2026; usage limits help verified Oct 24, 2026). Over-pool Cursor-model rates: Grok 4.7 $2 in / $6 out, Composer 2.5 $0.50 / $2.50 per million tokens.

Claude Ultra. At a limit you can turn on usage credits, switch to a Console account with API credits, or wait for the reset (verified Oct 23, 2026). Usage credits are billed “at standard API pricing rates” and “In most cases, usage credits do not expire” (usage credits article, verified Oct 21, 2026). Max also gets a small “wrapping up” allowance each time a five-hour limit hits mid-response. And if ANTHROPIC_API_KEY is set in your shell, Claude Code bills the API instead of your plan, so check that before blaming Anthropic.

Both sides end in the same place: overflow at API rates. The difference is what’s included before that. Cursor’s included pool is Grok and Composer; Claude’s is Claude.

What bugs me about both: at $200 a month, neither will say in plain numbers how much work you get. “50x Plus” and “10x Pro” are multiples of numbers nobody publishes. An exec who answered Logan Roy that way would be on all fours playing boar on the floor.

Who should pick which

Claude is your model, full stop: Claude Ultra. Running Claude Opus 5.5 all week inside Cursor’s own agent means paying Cursor’s $200 plus API list price for every token.
You live in an editor and run several agents, Automations or Bugbot reviews: Cursor Ultra. That’s what Cursor says Ultra is for.
You want Claude in Cursor without paying twice: use Cursor (even on a cheaper plan) as your editor and run Claude Code inside it on Max. Anthropic says IDE usage in Cursor counts toward your Max limits.
You want a yearly price: only Cursor sells one ($2,000).
You’re on legacy Cursor Ultra: your plan still includes a third-party pool. Switching is optional and permanent, so compare first.
You’re not sure you need $200: step down a tier first. Cursor Pro vs Pro+ and Claude Max vs Ultra cover that.

How the weekly pool behaves day to day is in Cursor usage limits. The entry-tier version of this fight is Cursor vs Claude Code.

And Codex Pro?

OpenAI’s $200 tier, Codex Pro $200, “includes more usage than Pro 100” and “Pro plans currently have no five-hour limit”, but OpenAI publishes no Pro usage figures and new subscribers get “a lower usage allowance” than grandfathered ones (Codex pricing, verified Oct 25, 2026; Pro plans article, verified Oct 21, 2026). The full breakdown is in Codex Pro vs Claude Code Max, and the wider field is on AI coding tools compared.

Here’s what I’d do

We haven’t run either plan for a month, so this is a read of the vendor pages, not a benchmark. If I wanted Claude, I’d buy Claude Ultra and run Claude Code in whatever editor I like, Cursor included. If I wanted Cursor’s agent stack and was happy on Grok and Composer, I’d buy Ultra yearly once I’d proved a month of it. What I wouldn’t do is buy Ultra to use Claude in it. That’s $200 for the wrong pool plus a per-token bill for the right one.

Also weighing ChatGPT? Claude Max vs ChatGPT Pro compares the two big chat subscriptions.

Comparing Cursor’s top plan with OpenAI’s instead? Cursor Ultra vs ChatGPT Pro covers the other $200 matchup.

On legacy Cursor Pro+ instead of Ultra? Cursor Pro+ vs Claude Max compares the $60 legacy plan with Claude Max at $100.

Coming from legacy Pro+ rather than Super? Cursor Pro+ vs Ultra covers who should pay $200.

FAQ

Is Cursor Ultra worth it over Claude Max? Only if you want Cursor’s own models and agent features; for Claude usage, Claude Ultra includes Claude while Ultra bills it at API list price.

Can you use Claude in Cursor Ultra? Yes, but Claude isn’t in Ultra’s included pool; it’s billed at Anthropic’s API list price from prepaid credit, or via your own API key.

Which has higher limits? Neither publishes absolute numbers; Ultra is “50x Plus” weekly for Cursor models and Claude Ultra is 10 times Pro every week with no five-hour limit for Claude.

Does Claude Max work inside Cursor? Yes, Claude Code can run in Cursor, and Anthropic says that IDE usage counts toward your Max limits.

Sources

Cursor, Models & Pricing (plans, Ultra description, weekly pool, third-party models at API list price, Claude model rates, overage, annual prices): https://cursor.com/docs/models-and-pricing (verified 2026-10-26)
Cursor, Pricing help (legacy plans, switching, dropped third-party allowance): https://cursor.com/help/account-and-billing/pricing (verified 2026-10-24)
Cursor, Usage limits help (prepaid credit expiry, limit behaviour): https://cursor.com/help/models-and-usage/usage-limits (verified 2026-10-24)
Cursor, API keys help (chat models only, provider billing): https://cursor.com/help/models-and-usage/api-keys (verified 2026-10-23)
Anthropic, Claude pricing (Max features, Fable share, priority access, API credits, usage FAQ): https://claude.com/pricing (verified 2026-10-26)
Anthropic, What is the Max plan? (price, 20x multiple, weekly limit, monthly only): https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-10-24)
Anthropic, Use Claude Code with your Pro or Max plan (shared limits, IDE usage, Fable limit, options at limit, wrap-up, API key override): https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-10-23)
Anthropic, Usage credits for paid Claude plans: https://support.claude.com/en/articles/12429409-extra-usage-for-paid-claude-plans (verified 2026-10-21)
OpenAI, Codex pricing: https://developers.openai.com/codex/pricing (verified 2026-10-25)
OpenAI Help Center, About ChatGPT Pro plans: https://help.openai.com/en/articles/9793128-about-chatgpt-pro-plans (verified 2026-10-21)