Cursor's $20 Plan Is Gone: X Subscription Pricing and Two November Deadlines

The $20 Cursor Plan Is Gone: What the X Subscription Actually Changed

On October 12, 2026, SpaceXAI — the AI unit formerly known as xAI — launched The X Subscription, collapsing Grok, the Grok Bot agent, the Cursor code editor, and X Premium into one price list: Free, Plus at $8, Premium at $30, Super at $100, Ultra at $200 per month. Cursor's pricing page now states that since that date, new Cursor plans are Plus, Premium, Super and Ultra — and the $20 Pro plan is closed to new signups.

Existing Pro, Pro+ and Ultra subscribers keep their plans, and their included usage, for now. Everyone else walks into a different product.

Cheaper, or just different?

Read as a price cut, it is one of the year's biggest. X Premium+ alone was $40, SuperGrok alone was $30, and Cursor Pro alone was $20 — roughly $90 combined. Premium at $30 is said to replace all three.

Read the fine print instead. From Cursor's own pricing documentation:

Plus ($8) — "Only Grok models are available on this plan." No cloud agents, no Bugbot, no bring-your-own API key.
Premium ($30) and above — third-party models are available in Cursor, "billed separately by usage."

Claude and GPT were not removed from Cursor. They were moved out of the subscription and onto a meter charged at API list price. The usage pool you pay $30, $100 or $200 a month for covers Grok models and Cursor's own Composer, and nothing else.

The published per-million-token rates tell the story:

Grok 4.6 — $2 input, $6 output, $0.50 cache read
Composer 2.5 — $0.50 input, $2.50 output ($3 / $15 in Fast mode)
Claude 4.5 Sonnet — $3 input, $15 output
Claude 4.6 Opus — $5 input, $25 output

One pool, two economics.

The quiet change: weekly resets and a shared meter

The old plans reset monthly. The new ones reset weekly. And the pool is shared with Grok chat, voice, image generation and Grok Bot — so generating images on X draws down the same allowance that powers your Cursor agents.

For individuals that is convenience. For teams it is a governance problem. "A shared usage pool makes AI coding spend harder to attribute, and a team that can't meter its coding agents can't govern them at scale," Mitch Ashley, VP at The Futurum Group, told DevOps.com. "SpaceX is pricing Cursor like a perk on a social media plan."

What existing subscribers should do

Cursor's message to current subscribers is that nothing changes automatically. If your daily work runs mostly on Claude or GPT, staying put is probably cheaper — old plans retain their included third-party allowance.

The migration is one-way, though. Switching permanently merges your Cursor account with a Grok and X account, and the old plan ends with a prorated refund. A lapsed subscription cannot be restarted, because the legacy plans are no longer sold.

Three realistic paths:

Keep the legacy plan. It is no longer purchasable, which makes it an appreciating asset until Cursor announces a forced migration.

Move to Premium and pay per token for Claude or GPT. $30 plus API-rate usage, with the option to bring your own key (Premium and above only — Plus cannot).

Prepare for the OpenAI cutoff. Separately but on the same calendar: OpenAI said on August 28 that it will wind down its model-supply contract to Cursor — now owned by SpaceX after a $60 billion all-stock acquisition that closed August 14 — with a proposed shutoff on November 12, 2026. Cursor CEO Michael Truell said OpenAI models serve about 5% of Cursor user traffic. If that 5% is the core of your workflow, the impact is 100% for you.

November 16 is the other deadline

GitHub's October 15 changelog adds a second date: Claude Opus 4.8, Claude Opus 5 and GPT-5.6 Terra are deprecated across all Copilot experiences on November 16, 2026, with Claude Opus 5.5 and GPT-6.1 Sol or GPT-6 Luna as suggested replacements. Individual annual-plan subscribers keep Opus 4.8 as an exception.

Two major coding tools, two model lineups changing within four days of each other. Any team that has not written down which model IDs it depends on will find out the hard way.

The product kept moving, too

The pricing story obscures a real product push. Cursor shipped three times in October:

October 6 — the iOS app gained remote control for agents running locally on your own machine, plus the ability to hand a local plan to a cloud agent and pull it back.
October 13 — Origin, an app marketplace, opened with 20+ integrations across deployments (Cloudflare, Netlify, Render, Spacelift, Vercel), CI (Buildkite, CircleCI, CloudBees, Depot, EngFlow, Namespace, RWX), code review (CodeRabbit, Greptile, Tenki), security (Artemis, Semgrep, Socket, Upwind), monitoring (Sentry) and project management (Linear) — with public submissions open from day one.
October 16 — cloud agents can build shareable interactive canvases; the Environments API entered beta; environments now support up to 300 repositories with 5x faster loading for large ones.

Cursor also publishes its own adoption figure: one in ten of its merged pull requests came from cloud agents last December; today it is 56%.

The decision framework

If Grok covers your work, Plus at $8 is genuinely cheap. An editor, Grok API access and X Premium for the price of a large coffee subscription.
If you live in Claude, measure before you switch. Record a week of actual token consumption, then price it at API rates against the new allowance. Guessing here is how teams double their spend.
If you run a team, demand attribution first. If you cannot separate coding usage from chat and image generation inside the shared pool, chargeback is broken on day one.

Bundles always win on features-per-dollar. They lose the moment you want to swap out one component. Decide whether you intend to leave in a year before you decide whether the discount is worth taking.