On this page
Cursor Pricing Explained
Cursor's June 2026 Teams pricing overhaul introduces split usage pools that make its proprietary Composer model far cheaper than third-party options like Claude and GPT. This structure is a deliberate lock-in strategy to push teams toward Cursor's full proprietary AI coding stack, not just a response to cost complaints. Individual plans use a credit pool system where Auto mode does not drain credits, making Pro plans sufficient for most developers.
Cursor crossed $2 billion in annualized revenue in early 2026, and it’s used by 64% of Fortune 500 companies. Those numbers tell you something important: this tool has moved well past the “promising startup” phase. But the pricing architecture underneath that growth has become genuinely complex, and the June 2026 Teams overhaul reveals a strategy that goes far beyond simplifying your bill.
The June 2026 Teams Overhaul: What Actually Changed
Cursor restructured its Teams plan in June 2026, introducing two separate usage pools per seat: one for Composer and Auto (first-party Cursor models) and one for third-party API models like Claude and GPT. The changes took effect immediately for new customers, with renewing customers on billing cycles starting July 1, 2026.
The new Teams structure offers two seat types:
- Standard: $32/seat/month annual, $40/seat/month monthly
- Premium: $96/seat/month annual, $120/seat/month monthly (5x the usage at 3x the cost)
Cursor expects the Composer usage pool on Premium seats to cover a full month of heavy agent usage for 99% of users. The dashboard now provides real-time usage visibility split between Composer/Auto and third-party API usage, with configurable spend alerts via Slack or email for admins.
On the surface, this looks like a response to the “unpredictable cost” complaints that plagued the August 2025 shift to token-rate billing. And it is, partially. But the architecture tells a different story when you look at where the economic incentives point.
The Vertical Stack Capture Pattern
What I call the Vertical Stack Capture pattern is playing out in real time. Cursor is executing a deliberate strategy to move from a thin orchestration layer over third-party AI models and GitHub to owning the full AI coding stack. The June 2026 pricing restructure isn’t a reaction to user complaints about unpredictable costs—it’s a pre-emptive lock-in strategy designed to make its proprietary Composer 2.5 model the default for 99% of routine team workflows, while gating third-party frontier models behind expensive Premium seats.
Consider the math. Composer 2.5 Standard runs at $0.50 per million input tokens and $2.50 per million output tokens per Technobezz. A Fast variant costs $1.50/$7.50. Cursor’s first-party model is roughly one-sixth the cost of frontier third-party alternatives.
The split usage pools make this economic gap visible to admins in real time. Teams see exactly how much more they’re spending when developers reach for Claude Opus instead of Composer. The natural administrative response—steer most users to the cheaper pool—aligns perfectly with Cursor’s strategic goal.
This pricing move coincided with Cursor’s Compile 2026 event, where the company announced Origin (a git hosting platform built for AI agents), Cursor Mobile, and a 1.5-trillion-parameter model being trained from scratch with SpaceX. Each piece reduces dependency on third-party infrastructure. The pricing architecture makes adopting that integrated stack the path of least resistance.
Individual Plans: The Credit Pool System
For individual developers, Cursor’s pricing starts at $0 for the free Hobby plan and tops out at $200/month for Ultra. The credit-based billing system, implemented in June 2025, allocates a monthly usage pool equal to the plan’s monthly price in dollars. This pool depletes only when users manually select premium models or enable Max Mode.
Here’s the full individual tier breakdown:
| Plan | Monthly Price | Annual Price | Credit Pool | Best For |
|---|---|---|---|---|
| Hobby | Free | — | Limited Agent + Tab | Evaluation |
| Pro | $20 | $16/month | $20 | Daily developers |
| Pro+ | $60 | $48/month | $70 | Heavy Agent users |
| Ultra | $200 | $160/month | $400 | Power users, all-day agents |
Auto mode is unlimited on all paid plans and does not consume the monthly credit pool. This is the detail most developers miss: if you stay in Auto mode, your $20 credit pool on Pro may last far longer than you expect. Credits only deplete when you manually select a specific frontier model or enable Max Mode.
Annual billing provides a 20% discount on all paid plans.
Notably, all individual tiers provide access to the same set of AI models. Higher tiers only increase usage limits, not model access. The upgrade decision is purely about whether you’re burning through your credit pool each month.
The Real Cost Math for Teams
Let’s put concrete numbers on the Teams pricing. A 10-person deployment with 2 Premium seats (for heavy agent users) and 8 Standard seats (for regular users), billed annually, costs approximately $448 per month in subscription fees. An equivalent all-Premium annual deployment runs $960 per month.
The math: (2 × $96) + (8 × $32) = $448 versus 10 × $96 = $960.
For most teams, assigning Premium to your 2-3 heaviest users and Standard to everyone else cuts the monthly bill nearly in half versus going all-Premium—and the other 8 developers won’t notice the difference in daily workflow.
But here’s the tradeoff you’re actually making. Cursor can offer lower per-seat costs for standard users by routing most usage to its low-cost proprietary Composer 2.5 model. This limits access to higher-capability third-party frontier models for complex tasks, which are now confined to smaller, more expensive Premium seat pools.
The Lock-in Problem Nobody’s Talking About
Cursor’s full-stack expansion—proprietary model, Origin git hosting, cloud agents, mobile control—creates a tightly integrated, low-latency workflow for users who adopt the entire stack. That’s the upside. The downside is severe vendor lock-in that makes switching to competing tools prohibitively expensive for teams that have built processes around Cursor’s proprietary tooling.
Once your code lives on Origin, your agents run on Cursor’s cloud infrastructure, and your team’s workflows are tuned to Composer 2.5’s capabilities, migrating to GitHub Copilot or Claude Code isn’t a matter of swapping editors. It’s a full infrastructure migration.
This is the contradiction at the heart of Cursor’s current positioning. The company markets itself as a model-agnostic tool that supports all major third-party frontier models. Multiple independent reviews highlight multi-model support for Claude, GPT, and Gemini as a core feature. But the product and pricing moves signal a deliberate shift toward a closed, proprietary stack.
The split usage pools and Premium seat pricing make third-party frontier models 3-5x more expensive per unit of usage than Composer 2.5, effectively forcing most users to adopt the first-party model to avoid overages. That’s not model agnosticism—that’s economic steering.
How This Compares to the Rest of the Market
Cursor’s pricing architecture now differs fundamentally from both GitHub Copilot and Claude Code. Cursor switched to usage-based billing in June 2025, while the June 2026 update restructured Teams pricing within that existing usage-based system. This creates different cost structures for engineering teams to navigate.
For a deeper breakdown of how these models compare, see our complete comparison of Copilot, Cursor, and Claude Code. If you’re specifically weighing Cursor against Claude Code’s agent capabilities, the agent mode comparison reveals a structural cost inversion behind their identical entry prices.
The key distinction: Claude Code offers flat per-seat pricing with zero overage risk. You know exactly what you’ll pay each month. Cursor’s split pools and flexible seat mixing give teams fine-grained control over spend and reduce overages for most users, but they add administrative overhead for team admins who must monitor usage across two separate pools and assign seat types to individual users based on their behavior.
What I’d Actually Recommend
For individual developers, start on Pro at $20/month. Use Auto mode as your default, manually select frontier models only when the task genuinely requires it, and track whether you’re consistently exceeding your $20 credit pool before upgrading. Most developers don’t need Ultra—start on Pro, upgrade to Pro+ if you’re regularly paying $20-40 in overages.
For teams, the mixed seat strategy is financially obvious but architecturally consequential. Assign Premium seats to your heaviest agent users and Standard seats to everyone else. But before you build your workflow around Composer 2.5 and Origin, ask whether you’re comfortable with the switching costs 12 months from now. If your team regularly needs frontier model capabilities that Composer 2.5 can’t match, the “savings” from the Standard seat pool may cost you more in capability than you save in dollars.
The question isn’t whether Cursor is worth the price. It’s whether you’re buying a tool or buying into a stack. Those are very different commitments.