Cursor’s New $30 Plan Includes $0 of Claude. Copilot’s $10 Plan Includes $15 in Credits.

What changed

Cursor is now sold as part of The X Subscription, which Cursor’s docs describe as “one subscription for Grok, Cursor, and X.” SpaceXSI’s own help page for the bundle carries a last-updated date of October 12, 2026. New customers start on the new plans. Existing customers keep their old ones unless they switch.

The new ladder copies the old price points almost exactly: Free, Plus at $8, Premium at $30, Super at $100 and Ultra at $200 a month. What moved is what the money buys.

Every paid plan includes one weekly pool of usage for “Cursor models”: Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5. That pool is shared with the Grok apps, Grok Bot and the CLI. Third-party models sit outside it. In Cursor’s words: “Third-party models like Claude, GPT, and Gemini aren’t part of the included pool.”

The key numbers

Source: Cursor’s Models & Pricing and Legacy Cursor plans docs, read October 25, 2026. Cursor maps legacy Pro ($20) to Premium ($30), Pro+ ($60) to Super ($100), and Ultra ($200) to Ultra ($200).

So the trade is explicit. More Grok and Composer, zero included Claude. If your workflow already lives on Composer or Grok, the new plans are a better deal by Cursor’s own description. If you opened Cursor to run Claude, they are not.

What a month of Claude costs on the new plans

Cursor publishes per-token rates for every third-party model it serves. Here are the three current Claude tiers, per million tokens:

Now a worked example. It is our illustration, not a Cursor figure: a month with 20 million cache-read tokens, 2 million fresh input tokens and 1 million output tokens. That is a modest month for someone running an agent on a real codebase. Cache writes are left out to keep the math simple, so the true bill would run a little higher.

The arithmetic for Sonnet 5.5: 20M × $0.10 = $2, plus 2M × $2 = $4, plus 1M × $10 = $10, for $16. Opus 5.5 doubles each rate, so $32. Fable 5.1 lands at $5 + $20 + $50 = $75.

Notice where the money goes. Output tokens are 1 million out of 23 million in this example, yet they make up $10 of the $16 Sonnet bill. Agents that write long diffs and long explanations pay for it on output, not context.

Teams pay a toll on top

On the surface, teams got a price cut. The legacy Teams Standard seat was $40 per user. Its mapped replacement, a Premium seat, is $30. That is 25% less.

But team seats also lose the included third-party allowance, and they pick up a new fee. On Teams and Enterprise, eligible third-party requests carry a “Cursor Token Rate of $0.25 per million tokens,” applied to input, output and cached tokens alike. Individual plans don’t pay it.

Run our 23-million-token example through a team seat on Claude Sonnet 5.5 and the token rate adds 23 × $0.25 = $5.75 to that seat’s month. Cheap models such as Claude Haiku 5.5 and GPT-5.6 Luna are on Cursor’s exclusion list and skip the rate.

How the alternatives stack up

GitHub defines 1 AI credit as $0.01, so Copilot Pro’s $15 is $15 of model usage for $10. GitHub also warns that “flex allotments may change over time,” so treat the extra $5 as soft. Code completions don’t use credits at all.

The comparison is not perfect. Cursor is a full editor with cloud agents, automations, code hosting and AI code review from Premium up. Copilot and Claude Code price their usage differently. But on the narrow question of “how much Claude does my subscription include,” Cursor’s new answer is the lowest of the three: none.

What this means if you code with AI

If you’re on legacy Cursor Pro, don’t click Unify by reflex. Cursor’s help page says the switch is permanent: “It can’t be undone, and Support can’t undo it for you.” Check your usage dashboard first. If most of your spend lands in the Other Models pool, the new plans will cost you more.

If you mostly use Composer, Grok or Auto, the new plans can be a win. Cursor says every new tier carries at least twice the Cursor-model usage of the plan it replaces, and the pool refreshes weekly instead of monthly.

If you’re new and want Claude, budget for the seat plus the tokens. The $30 Premium seat is a floor, not a ceiling. Set a monthly additional usage limit on day one, since prepaid credit pays for Claude “from the first request.”

Watch output-heavy work. At $10 to $50 per million output tokens, long agent runs that rewrite whole files cost far more than the context they read.

Separate the model from the editor. The cheapest path to a given model now depends on who bundles it. A Claude Sonnet 5.5 token billed at list price costs the same in any tool; what differs is how much usage your subscription covers before that meter starts. On Cursor’s new plans, for Claude, that number is zero.

Where MegaBrain fits

Full disclosure: MegaBrain uses the same plan ladder, $8, $30, $100 and $200 a month. On MegaBrain’s $8 Plus plan the usage covers open models (GLM, Kimi, MiniMax, DeepSeek, Qwen), Grok and Auto, not Grok alone. Claude, GPT and Gemini come from credits from Pro, and cost more per token than they do on Cursor. Your own API keys and local models are free on every plan, Free included, and never count against a limit.

If that trade fits how you work, the side-by-side is on our Cursor alternative page, and every plan is on the pricing page. MegaBrain is not affiliated with Cursor.

Sources

All pages opened October 25, 2026.

• Cursor — Pricing (plans, Grok-only Plus, third-party models billed separately)
• Cursor Docs — Models & Pricing (per-model token rates, Cursor Token Rate)
• Cursor Docs — Legacy Cursor plans (Pro, Pro+, Teams mapping)
• Cursor Help — What is The X Subscription? (the switch is permanent)
• SpaceXSI Docs — What is The X Subscription? (last updated October 12, 2026)
• SpaceXSI Docs — The X Subscription pricing (annual prices, team seats)
• GitHub — Copilot plans (Pro $10, $15 in AI credits, 1 credit = $0.01)
• Anthropic — Claude pricing (Pro $20, Claude Code included)