nicota

From the email update:

We're updating our Teams pricing to better adapt to an agentic world. These changes will start at your next billing renewal after September 15 for both monthly and yearly plans.

From fixed to API-based usage

Today, our Teams plans charge a fixed-cost per request and include the following:

Unlimited Tab and access to all core features 250 Sonnet requests per month Additional usage is billed at $0.08/Sonnet request As models have become more agentic, the cost of individual agent requests has grown increasingly variable. Using the same model, a single difficult prompt can consume an order of magnitude more tokens than a simple one.

Fixed costs per request aren't suited to this new reality. They force an overpricing of simple requests (e.g. a quick syntax question) and an underpricing of difficult ones (e.g. asking the agent to complete an entire PR). API-based pricing will better align both with the underlying LLM costs and with the value the agent delivers.

Under the new pricing, our Teams plan will transition to the following:

Unlimited Tab and access to all features $20 of agent usage per user Ability to purchase additional usage beyond this All usage is consumed at public list API prices of the underlying model + $0.25 per million total tokens. This additional cost covers Cursor's code indexing, indexing model costs, and tool execution with custom models.

This change unifies our billing system. Our individual plans already use API-based usage in production.

What's next

Pricing changes will start at your next billing renewal after September 15. If you need pooled usage or want to explore volume discounts, please reach out to [email protected]. We're grateful for your feedback as we continue evolving Cursor to better support your needs.

Interesting! May I ask how many employees did it serve? Did you have to compile large C++ projects and noticed any performance issues? Were the cores shared among different users? (overbooking the number of cores)

Thanks!

That seems quite good, but I'm wondering about scalability - it's not a single desktop, but rather around ~100 employees that need serving...