Orca Pod — a coding plan for any client
A coding plan for people who run agents, not for people who integrate an API. Limits are in agent-hours, never prompt counts, and your price is locked for 12 months.
Three windows, whichever you reach first
Usage is metered over a 5-hour, a weekly and a monthly window, each stated in agent-hours: about how long an agent can work on the router. How far the hours go depends on which models the router picks.
Locked for 12 months
Your price is locked for 12 months from your first full-price payment, so a later change to the published plan does not move what you pay.
Four questions
What happens when I hit a limit?
Requests pause until the reset shown on your plan page — or switch on Overflow and keep going from your wallet at published prices, zero markup. Overflow is off by default.
Do I have to use the router?
No. orcarouter/pod picks the model per turn — routine turns on the open-weight pool, hard turns on frontier — but naming a model from the plan list directly works too.
Do you store my prompts?
Not by default. Plan-funded requests are exempt from payload capture, so nothing you send is kept unless a workspace admin turns request details on in your plan settings. Overflow requests are paid by the wallet, not the plan, and are captured under your normal workspace setting. Billing rows (model, tokens, cost) stay in your console logs either way.
I have Team. Do I need Pod too?
No. Team includes one Pod allowance for the whole workspace, pooled across members through any workspace key.