One price, however busy the day gets

Questions

What counts as a request?

One call to your deployed endpoint: a visitor asking your assistant one thing. Requests your assistant answers from its own training cost the same as any other.

What is an AI fallback?

When your model does not recognise a question, we can hand it to a general-purpose model so the visitor still gets an answer. That call costs real money, so it is metered separately. Free plans have none; the assistant says it does not know instead.

What does “unlimited, subject to fair use” mean?

Pro has no per-request charge and no practical daily ceiling. There is a fair-use cap high above normal usage to stop one account degrading the service for everyone. Your profile shows what you have used rather than counting you down toward it, and we watch the ceiling so you do not have to. If you expect to approach it, talk to us first and we will size a plan around it.

What happens if I hit a limit?

The endpoint answers with a quota message rather than failing silently, and the limit resets at 00:00 UTC. We email you at 80% and again at 100% of a metered limit, and you can ask us for grace credits to finish the day. Unlimited requests get no 80% warning, because there is no allowance to spend down.

Can I cancel?

Yes, from your profile, at any time. You keep your plan until the end of the period you have paid for, and your projects, models and endpoints stay where they are.

Do you offer a trial?

Pro starts with a free week. You are not charged until it ends, and cancelling during it costs nothing.