Cursor Pricing 2026: Is It Free, and What You Pay
Cursor has a $0 Free plan with limited agent and chat on its own models. Real use now starts at Premium, $30/month — not the $20 Pro plan, which closed to new sign-ups on 12 October 2026. The $8 Plus tier runs Grok models only, and Claude, GPT and Gemini bill separately at API list price on Premium and above. Teams dropped to $30 per seat on 19 October.
AICoding score
Quick Verdict
8.7 / 10 — Still the best AI editor, now sold as a bundle you have to unpack.
Cursor is the editor that made AI-native editing normal, and the agent and Composer experience is still the benchmark others copy. What changed in October 2026 is how you pay for it: the price list belongs to The X Subscription, the usage pool is shared with Grok chat, the cheapest tier is limited to one vendor's models, and the models most developers actually want are metered at API list price on top. Read the invoice in two parts — Cursor capacity, frontier model tokens — and set a spending cap. Treat Premium as the subscription and Super or Ultra as capacity purchases rather than feature upgrades.
Cursor is right if…
- You want the deepest agent and multi-file editing UX in one editor
- You are prepared to budget $30/month plus third-party model usage
- You need Composer-style whole-project edits rather than autocomplete
- You want a team rollout with SSO, analytics and shared rules at $30/seat
Look elsewhere if…
- You need a predictable flat bill — look at Copilot Pro at $10 or a BYO-key tool
- You want frontier models included in the subscription, as they were under Pro
- You are a student hoping for the old discount, which closed in June 2026
- Your usage is mostly autocomplete, which other tools do not meter
What You Pay For
Agent Mode
Cursor's agent works across multiple files, runs commands and iterates toward a goal inside the editor. It is the feature most competing editors are measured against, and the main reason developers pay $30 rather than $10.
Composer
Whole-project editing from a single prompt: describe the change, review a multi-file diff, accept or reject per hunk. Available even on the free plan, which is unusual for a headline feature.
Model Choice, Two Wallets
Grok models draw on the plan usage pool; Claude, GPT and Gemini bill at API list price outside it, with an optional spending cap. Picking a model is now a pricing decision as much as a quality one.
Bugbot Code Review
Agent-driven pull-request review, billed by usage on personal plans and included in Teams. It puts Cursor in the same category as dedicated review tools rather than leaving review to a separate vendor.
Teams Governance
Centralised billing, a shared marketplace for rules, skills and plugins, cloud agents with team context, usage analytics, team privacy mode and SAML/OIDC SSO at $30 per user per month.
Privacy Mode
Guarantees that neither Cursor nor its model providers train on your code, configurable per user or enforced by a team administrator — the feature enterprise buyers ask about first.
Pricing
Limited agent and chat requests on Cursor's own models, no card required. No access to Claude, GPT or Gemini. Enough to evaluate the editor, not to run it daily.
- Limited agent requests
- Cursor models only
- No card required
- Evaluation only
The 1x usage baseline — and Grok models only. This is the tier most reviews do not explain: it is a cheap entry point to the bundle, not to the editor with frontier models.
- 1x usage pool
- Grok models only
- Weekly pool refresh
- No third-party models
The practical starting point and the plan Cursor labels most popular: 4x Plus usage, cloud agents and automations, code hosting, Bugbot review, and third-party models at API list price outside the pool.
- 4x Plus usage
- Cloud agents + automations
- Bugbot review
- Third-party models at API price
Capacity, not features: 15x and 50x the Plus pool, with priority access at peak times on Super and earliest access to new models on Ultra. Annual billing is $1,000 and $2,000.
- 15x / 50x usage
- Priority or earliest access
- Same feature set
- Annual: $1,000 / $2,000
Down from $40 on 19 October 2026. Seat tiers run $30, $100 and $200 depending on per-seat usage. Centralised billing and admin, shared team marketplace, cloud agents with shared context, analytics, team privacy mode and SAML/OIDC SSO.
- Centralised billing
- Team marketplace
- Usage analytics
- SAML/OIDC SSO
Invoice and purchase-order billing, SCIM seat management, group-level spend controls, repository/model/MCP access controls, audit logs, service accounts and priority support.
- SCIM + audit logs
- Spend controls
- Invoice billing
- Priority support
* Plans changed on 12 October 2026: Free / Plus $8 / Premium $30 / Super $100 / Ultra $200, Teams $30+ per seat (verified 24 October 2026 on cursor.com/pricing). Pro ($20), Pro+ ($60) and Teams Standard ($40) are closed to new sign-ups and remain only for existing subscribers, who keep a separate third-party model pool. Claude, GPT and Gemini usage is billed at API list price outside the plan pool. Cursor publishes no fixed included-usage amounts.
Cursor Real Cost Breakdown
| Task / Plan | Usage | Typical cost | What to watch |
|---|---|---|---|
| Free plan | Limited agent and chat, Cursor models | $0 | Enough to test the editor, not enough to run it daily — and no frontier third-party models at all. |
| Premium subscription | $30/mo including a weekly Grok-model pool | $30/mo | The tier that actually includes the coding features. Included amounts are still not published. |
| Third-party models | Claude, GPT or Gemini selected in Cursor | API list price | Billed per use outside the pool on Premium and above. This is the line that turns $30 into $80 in a heavy month. |
| Shared pool spillover | Heavy Grok chat, Imagine or voice in the same week | Top-up or wait for refresh | The pool is shared across Cursor and the Grok products and refreshes weekly, so one product can starve the other. |
| Agent-heavy month | Long multi-file sessions on frontier models | $50–150 total | Subscription plus API-rate tokens. Set a spending limit in settings before the first month, not after. |
| Teams rollout | 5 seats at $30 + third-party usage | From $150/mo | Down from $200/mo at the old $40 seat price, but model spend still sits on top. |
| Annual billing | Prepay 12 months | $80 / $300 / $1,000 / $2,000 | About two months free versus monthly, for Plus through Ultra. There is no annual Teams discount listed. |
* Sourced from cursor.com/pricing (verified 24 October 2026) and the 12 October 2026 plan change, with the Teams seat price from 19 October 2026. Included usage amounts are not published, and third-party model spend is billed at the vendor rate, so the only honest way to budget is to measure your first month with a spending cap in place.
Cursor Deals & Official Discounts
There is no trial and no student discount for new sign-ups — the education offer closed on 25 June 2026. Cursor authorises no resellers; annual billing is the only official way to pay less.
Cursor vs The Field
Cursor vs GitHub Copilot
Depth versus breadth and priceCopilot Pro is $10, works across ten editors and does not meter completions; Copilot Pro+ is $39. Cursor is $30 for the tier that includes third-party models, exists only as its own editor, and bills Claude, GPT and Gemini on top at API price. Cursor's agent and Composer experience is deeper, which is what the premium buys. Completions-first work is cheaper on Copilot; agent-first work with frontier models is now a closer call than the old $20 versus $10 split suggested.
Copilot for cost, Cursor for depthCursor vs Windsurf / Devin Desktop
Two AI-native editors, different lineageWindsurf became Devin Desktop after Cognition acquired it, and it keeps its own free tier with editor-style assistance plus credit-based cloud agents. Both are VS Code-derived AI editors; Cursor has the deeper agent UX and larger ecosystem, Devin Desktop has the cloud-agent integration with Cognition's stack.
Cursor for agentsCursor vs Claude Code
Editor experience vs terminal reasoningClaude Code is a terminal agent with excellent long-horizon reasoning, priced via Anthropic subscriptions or API usage; it has no editor, no completions and no GUI. Many developers run both: Cursor for editing and Composer, Claude Code for the hardest refactors. Note that using Claude models inside Cursor now bills at API rates on top of the plan, so running both is a real cost decision rather than a free combination.
Frequently complementaryCursor vs Cline / Aider / Continue
Managed convenience vs BYO-key cost controlThese tools are free or open source with your own API key, no per-seat fee and no vendor markup on model usage. What you lose is Cursor's polish, the indexing quality and team features. Since Cursor now bills third-party models at list price anyway, a developer who only wants frontier models in an editor is comparing Cursor's $30 platform fee against a free harness.
BYO keys for cost controlWho Cursor Is Best For
Agent-first development
Multi-file tasks where you want the agent to plan, edit and iterate inside the editor with strong context handling.
Whole-project refactors
Composer handles coordinated changes across many files with per-hunk accept/reject — available even on the free plan.
Solo developers and freelancers
Budget $30 plus model tokens, set a spending cap, and measure the first month before deciding whether Super is worth it.
Teams standardising on one editor
Shared rules and plugins through the team marketplace, usage analytics for spotting waste, and SSO for procurement — now $30 a seat.
Regulated organisations
Enterprise adds SCIM, audit logs, group-level spend controls, model and MCP access controls, and privacy mode enforced centrally.
Budget-first users
Plus at $8 if Grok models are enough for the work; otherwise a BYO-key harness with your own model spend is cheaper than Premium.
Set Cursor Up Without a Bill Shock
Check which plan you are on
Existing Pro subscribers keep $20 and the old third-party pool; new sign-ups start at Plus $8 or Premium $30. Switching is optional and permanent, so read the terms before you click.
Start on Premium, not Plus
Plus is $8 but Grok-only. If you need Claude, GPT or Gemini, Premium is the real entry point and the extra $22 is buying access, not just capacity.
Set a spending limit on day one
Third-party models bill at API list price outside the pool. Set the cap in settings and watch the usage dashboard weekly for the first month.
Watch the shared pool
The weekly pool is shared with Grok chat, Imagine and voice. If you use the bundle for anything else, Cursor gets less of it than the plan implies.
Only then consider Teams
At $30 a seat it is cheaper than the old $40, but seat tiers go to $100 and $200 — pick per-seat sizes deliberately rather than defaulting everyone to the top tier.
FAQ
Is Cursor free?
There is a $0 Free plan, and it is narrower than the old Hobby tier: a limited number of agent and chat requests on Cursor's own models, with no access to Claude, GPT or Gemini. It is enough to evaluate the editor and not enough to run it daily. Real work now starts at Premium, $30 per month — the $20 Pro plan is closed to new sign-ups since 12 October 2026.
How much does Cursor cost in 2026?
Since 12 October 2026 the plans are The X Subscription: Free at $0, Plus at $8, Premium at $30, Super at $100 and Ultra at $200 per month, Teams from $30 per user, Enterprise by quote. Plus runs Grok models only. Premium and above include a weekly-refreshing usage pool for Grok models, and Claude, GPT or Gemini are billed separately at API list price on top. Annual billing is $80, $300, $1,000 and $2,000 respectively.
What happened to the $20 Pro plan?
It is legacy, not withdrawn. Cursor closed Pro ($20), Pro+ ($60) and Teams Standard ($40) to new sign-ups on 12 October 2026; existing subscribers keep their terms, including the separate third-party model pool. Cursor maps Pro to Premium at $30 rather than to Plus at $8, because Plus is Grok-only. Switching is optional and permanent, and there is no migration deadline — which is why two customers on the same editor can now be on two different price lists.
What is included at each tier?
Plus is the 1x usage baseline and runs Grok models only. Premium is 4x Plus and adds cloud agents and automations, code hosting (Origin), Bugbot review and the ability to bill third-party models at API list price. Super is 15x with priority access at peak times, Ultra is 50x with earliest access to new models. The pool refreshes weekly and is shared with Grok chat, Imagine and voice, so a heavy week in one product eats the other.
Are Claude, GPT and Gemini included?
Not in the plan. On Premium and above they are billed at API list price per use, outside the usage pool, with an optional spending cap. That is the single biggest change for anyone comparing this to the old Pro plan: the subscription now buys Cursor plus Grok capacity, and frontier third-party models are a separate line item at the same rate you would pay Anthropic or OpenAI directly.
How much did Teams change?
It fell from $40 to $30 per seat on 19 October 2026 and gained seat tiers at $30, $100 and $200 depending on how much usage each seat needs. Centralised billing, the team marketplace for shared rules and plugins, cloud agents with team context, usage analytics, team privacy mode, SAML/OIDC SSO and Bugbot are all in the Teams plan. A five-person team starts at $150 per month before third-party model usage.
Why is my Cursor bill higher than my subscription?
Because subscription and usage are separate. Third-party model calls are billed at API list price on top of the plan, and long agent sessions on frontier models, large codebase indexing and heavy Composer use across many files are the usual culprits. Set a spending limit in settings before you find out at month end, and check which models your requests are actually routed to.
Is there a free trial or a student discount?
Not for new sign-ups. The education discount closed on 25 June 2026, and Cursor's pricing page no longer lists a paid-plan trial — the 14-day Pro trial that most reviews still mention is gone. Anything advertising a cut-price Cursor subscription is not an authorised seller: Cursor sells only through cursor.com and states that subscriptions bought elsewhere may be suspended.
Is Cursor worth it versus GitHub Copilot?
The price gap widened. Copilot Pro is $10 per month, works across ten editors, and does not meter completions. Cursor is $30 for the tier that includes third-party models, and only exists as its own editor, with frontier models billed on top. The agent experience, Composer and multi-file editing are still deeper, which is what the premium buys. If your work is completions-first, Copilot is better value; if it is agent-first and you want frontier models, compare Cursor's $30 plus API usage against Copilot Pro+ at $39.
Does Cursor train on my code?
Privacy mode guarantees that neither Cursor nor its model providers use your code data for training, and it can be enabled per user or enforced by a team administrator. It matters most on personal plans, since organisational deployments generally control this centrally.
What are Cursor's main limitations in 2026?
Pricing that now has to be read as a bundle: one pool shared with a social product, an entry tier limited to one vendor's models, and third-party models metered separately. Add a single-editor commitment, no trial for new users, and two coexisting price lists that make comparison difficult. The editor quality is not the issue; working out what a given month costs before you start is.