Questions, answered.
Pricing, security, how it works, and how teams use Sweet.
How much does Sweet cost?
Free to start (usage-capped, no card). Pro $20/mo, Max $200/mo, and Team $1,000/mo for engineering teams (5–50 concurrent agents, unlimited seats).
What is the Team plan?
5–50 autonomous agents, unlimited seats, an admin + usage dashboard, and onboarding, for $1,000/mo — about 8% of one senior engineer. There's also a 30-day paid pilot.
How does Sweet work?
You give it a goal in plain English. It reads your repo, plans, writes the code, runs your tests, and opens a PR-ready diff you review like any PR.
Where does my code run?
The CLI runs locally on your machine — git, tests, and your tools execute there. The web app runs each session in its own isolated Firecracker microVM on Sweet's infrastructure.
Is my code sent to a model?
Yes — like any AI coding tool, the task context is sent to the model for inference. We don't build separate copies of your repo on shared storage as a service.
We already use Cursor/Copilot.
Different job. Those autocomplete while you drive; Sweet ships whole tasks end-to-end while you're away. Complementary.
Is it safe to let an agent run code?
Web sessions run in unprivileged, egress-restricted microVMs — one per session, isolated from other users and from our internal systems. The CLI runs locally.
Can we self-host?
Private deployment (agents in your own cloud) is on our Enterprise roadmap. Contact us.
How do we get started?
Start free at /billing.html, or book a 15-minute walkthrough on one of your real issues at /contact.html.
Ready?
Start free, or put agents on your team backlog.