As of July 31, 2026, Cursor Pro costs $20 / month on monthly billing, or roughly $16 / month on annual billing ($192 / year — about a 20% discount). That is the number most people want when they search "how much is Cursor Pro." But Cursor pricing in 2026 is no longer a simple "fixed monthly fee + unlimited requests" model. Since mid-2025, Cursor has shifted to credit-based billing: each plan includes a model usage pool equal to its price, Auto mode is effectively unlimited, and manually selecting frontier models consumes credits from your pool.
Bottom line: If you code daily and rely mainly on Tab completions and Auto mode, the $20 Pro plan is usually enough. If you run Agents all day and manually lock in Claude Opus or flagship GPT models, your real cost may land at Pro+ ($60 / month) or even Ultra ($200 / month). For teams, Teams Standard starts at $40 / user / month.
2026 Cursor plan pricing at a glance
Cursor's official pricing page (cursor.com/pricing) splits the product into four individual tiers, two team tiers, and custom enterprise pricing. The table below summarizes public list prices as of July 2026. Annual billing gives roughly a 20% discount off monthly rates. Prices exclude applicable local taxes.
| Plan | Monthly (USD) | Annual equivalent / mo | Included usage pool | Best for |
|---|---|---|---|---|
| Hobby (free) | $0 | — | Limited Agent + Tab | Trying Cursor, light personal use |
| Pro | $20 | ~$16 | $20 credit pool | Everyday individual development (most popular) |
| Pro+ | $60 | ~$48 | 3× Pro pool (~$60) | Heavy Agent users |
| Ultra | $200 | ~$160 | 20× Pro pool (~$400) | All-day Agent orchestration, parallel tasks |
| Teams Standard | $40 / user | ~$32 / user | Pro-equivalent usage per seat | Small teams needing admin and SSO |
| Teams Premium | $120 / user | ~$96 / user | 5× Standard seat usage | Heavy Agent users within a team |
| Enterprise | Custom | Annual only (typical) | Pooled usage + advanced governance | 50+ developers, compliance and audit needs |
How credits work: Auto vs manual model selection
The key to understanding 2026 Cursor pricing is separating the subscription list price from your actual usage cost. Starting in June 2025, Cursor moved from fixed request counts to a credit model — which is also why many users feel "Pro got more expensive" or "the pool runs out too fast."
Three core rules
- Included credits ≈ monthly fee. Pay $20 for Pro and you get roughly a $20 model usage pool. Pro+ gets ~$60, Ultra ~$400 (20× Pro).
- Auto mode does not draw from the pool. Conversations and Agents where Cursor picks the model automatically, plus Tab-related Auto usage, typically do not count against your credit pool on paid plans — the main reason Pro still feels "enough" for everyday users.
- Manually selecting flagship models draws from the pool. When you lock in Claude Sonnet / Opus, GPT-4, or other frontier models from the model picker, usage is deducted from your monthly pool at API-equivalent rates. After the pool is exhausted, you can enable on-demand usage billed at API prices with no markup.
How many requests your pool supports depends on model pricing and context size per request. Industry estimates suggest Pro's $20 pool roughly equals 200+ Claude Sonnet-class requests or 500+ mid-tier GPT requests — but a large Agent task with full repo context can consume several dollars in a single run. So "running Opus Agents manually every day" and "mostly coding with Tab + Auto" produce very different bills.
Cursor Pro in depth: what $20 actually includes
Pro is Cursor's best-selling tier. Compared to free Hobby, Pro unlocks:
- Unlimited Tab completions — AI inline completions in the editor without Hobby's strict limits.
- Expanded Agent limits — more Agent sessions, longer task chains, and Cloud Agent background execution.
- Full frontier model access — Claude, GPT, Gemini, Grok, and Cursor's own Composer model.
- MCP, Skills, and Hooks — connect external tools, custom skills, and workflow hooks.
- Maximum context window — include larger portions of your codebase in a single inference.
- Unlimited Auto mode — Auto model selection for chat and Agents does not consume your credit pool (manual selection excluded).
- ~$20 monthly credit pool — for premium requests when you manually choose high-cost models.
Bugbot (AI code review) is usage-billed on Pro and is not included in the $20 base subscription. If your team routes PR review through Cursor, track Bugbot usage separately.
Is Pro worth $20 / month?
For developers who code at least three days a week, Pro typically pays for itself within the first week through Tab completions and Inline Edit — time saved often far exceeds $20. If you code occasionally, the free Hobby tier or the cheaper GitHub Copilot individual plan (~$10 / month) may be a better fit.
If you run Agents all day locked to the most expensive models, Pro's pool will likely fall short — budget for Pro+ or enable on-demand billing with a hard cap. Honestly assessing your model usage habits matters more than comparing list prices alone.
Pro+ and Ultra: which tier heavy users should pick
| Comparison | Pro ($20) | Pro+ ($60) | Ultra ($200) |
|---|---|---|---|
| Credit pool | ~$20 / mo | ~$60 / mo (3×) | ~$400 / mo (20×) |
| Multiplier vs Pro | Baseline | 3× usage | 20× usage |
| Annual equivalent / mo | ~$16 | ~$48 | ~$160 |
| Official recommended use | Everyday individual workhorse | Daily Agent users | Agent power users, multi-agent parallelism |
Pro+ and Pro share nearly identical feature lists — the difference is mainly pool size. Upgrading from Pro to Pro+ adds $40 / month (3× the price) and 3× the credits — a good fit if you consistently exhaust the Pro pool and want to avoid frequent on-demand charges.
Ultra targets developers who treat Agents as their default workflow. Cursor 3.x supports parallel multi-agent runs (official docs mention up to 8 concurrent Agents). Long-running background tasks, large-repo refactors, and continuous Agent orchestration burn through credits fast. Ultra's $400 pool is designed for people who use AI as their primary development method, not for occasional Agent tasks.
Teams and Enterprise: buying for your organization
Team plans differ from individual plans in more than seat count — they add administration, governance, and collaboration capabilities.
Teams Standard ($40 / user / month)
Everything in Individual plans, plus:
- Centralized billing and seat management
- Team-wide Privacy Mode (code not used for training)
- SAML / OIDC single sign-on
- Usage analytics and Admin Dashboard
- Team Marketplace (shared Rules, Skills, plugins)
- Bugbot agentic code review
- Cloud Agents and automation with shared context
Each seat includes Pro-equivalent usage. A 5-person team on monthly billing runs about $200 / month (~$160 / month on annual billing).
Teams Premium ($120 / user / month)
Around June 2026, Cursor introduced Premium seats: 3× the Standard price with 5× the Standard usage. Ideal for the few "Agent power users" on a team — rather than giving everyone an Ultra individual plan, assign Premium seats to heavy users and Standard to the rest.
Enterprise (custom pricing)
For large organizations, Enterprise adds to Teams:
- Pooled usage — shared credits across the org to avoid wasted per-seat allocation
- Invoice / PO billing, SCIM seat automation
- Repository, model, and MCP access controls
- Auto-run, browser, and network policy controls
- Audit logs, service accounts, AI code tracking API
- Priority support and dedicated customer success
Typically aimed at teams of 50+ developers with compliance, audit, or industry regulatory requirements. Contact sales for a quote.
Free Hobby tier and Pro trial
Hobby is free forever with no credit card required, but Agent requests and Tab completions have clear limits — enough to evaluate whether Cursor fits your workflow.
New Hobby sign-ups generally receive a two-week full Pro trial. After the trial, you revert to Hobby limits automatically — no charge unless you subscribe.
One note: the free one-year Pro for students program stopped accepting new registrations on June 25, 2026. Existing approved students may retain remaining benefits, but new student accounts must purchase at standard pricing.
Price comparison with GitHub Copilot
GitHub Copilot is the most common alternative. As of July 2026, the two products use different billing philosophies:
| Dimension | Cursor Pro | GitHub Copilot Pro |
|---|---|---|
| Individual monthly price | $20 / month | ~$10 / month (Copilot Pro) |
| Product form | Standalone AI IDE (VS Code fork) | Editor extension + Copilot App |
| Billing model | Subscription + credit pool + on-demand | Subscription + AI Credits (since June 2026) |
| Completions | Unlimited Tab on Pro | Completions do not consume AI Credits |
| Agent capabilities | In-editor + Background / Cloud Agent | Copilot App multi-session + cloud sandbox |
| Team starting price | $40 / user / month (Teams Standard) | ~$19 / user / month (Business) |
Copilot's individual plan is cheaper, but Cursor delivers a full IDE experience with stronger Agent orchestration. If you already code in Cursor, $20 is often less hassle than "Copilot extension + original editor." For a deeper tool comparison, see our article GitHub Copilot App vs Cursor: Should You Switch in 2026?
Plan recommendations: which tier fits you
| Your situation | Recommended plan | Why |
|---|---|---|
| Occasional coding, just exploring | Hobby (free) | Zero-cost evaluation; upgrade to Pro when ready |
| Everyday individual dev, Tab + Auto primary | Pro (annual saves more) | $20 covers most scenarios — best value |
| Daily multi-round Agents, often manually picking flagship models | Pro+ or Pro + on-demand cap | Avoid hitting the pool mid-month and switching to Auto |
| Full-time AI-native development, parallel Agents | Ultra | $400 pool + unlimited Auto reduces billing anxiety |
| 5–20 person engineering team needing SSO and governance | Teams Standard | Admin panel + privacy mode + centralized billing |
| Few heavy Agent users within a team | Standard + Premium mixed seats | More economical than Ultra for everyone |
| 50+ people, compliance audit, pooled usage | Enterprise | Custom contract and advanced security controls |
Cost-saving tips and spending guardrails
- Annual billing saves 20%. If you plan to use Cursor long-term, Pro annual at $192 beats $240 on monthly billing — $48 saved.
- Prefer Auto; switch models deliberately. Use Auto for everyday chat and Agents; manually pick a specific model only when you need its capabilities.
- Set a spend limit. Enable Spend Limit in Cursor account settings to prevent on-demand usage from stacking surprise charges at month-end.
- Monitor Background Agents separately. Parallel background tasks amplify total model and environment costs — pilot with small batches before scaling up.
- Use mixed team seats. Do not give everyone Premium; assign Standard and Premium based on actual Agent usage.
- Factor in environment costs. Cursor subscriptions do not include remote Macs, CI minutes, or cloud hosts. Apple platform developers needing stable Xcode environments should budget for remote Mac rental separately (see below).
One point that's easy to overlook: Cursor is an editor and Agent platform, not a complete dev machine. For iOS builds, Xcode signing, or long local compile jobs, you still need a reliable macOS environment. If local hardware is insufficient or you need a 24/7 build node, run Cursor on a remote Mac so Agents and builds share the same stable environment.
Run Cursor on a remote Mac — factor environment costs into your budget
Cursor Pro's $20 covers AI capabilities, not compute and system environment. Apple platform development, parallel multi-Agent builds, or teams needing a fixed Xcode version often require an always-available Mac — upgrading or buying a Mac mini is a one-time hardware cost; cloud rental is predictable monthly spend.
Kvmjet offers Mac mini M4 cloud nodes with direct remote desktop access, fixed Xcode environments, and multi-region node selection. Install Cursor on a cloud Mac and let Agents edit code, compile, and sign in the same environment — avoiding the split where "the Linux Agent finished, but it won't build on macOS."