The lab is a phone line to the run engine that the console and the SDKs also use. It never prices, holds or refunds anything itself — it shows you what the engine decided.
Machines
The Run rail lists the catalogue: simulators and real hardware, from several providers behind one set of names. Each machine has
- a display name and a codename in mono — the codename is the public
targetId, the exact stringrun()/submit_run()and/v1/machinesuse. Same id in every SDK and the API; copy it with the button beside it. - a status chip: available (green), queue N (amber — jobs ahead of you; a machine that is currently failing health checks also reads busy), offline (grey; listed so the codename stays real, but disabled in the picker).
- simulator or real hardware, and the qubit count.
- in the reference list, a catalogue per-shot rate — list price for people writing code. Orgs can carry negotiated pricing, so the only number that binds is the quote at submit.
Simulators are cheap and instantaneous. Real machines queue for minutes to hours (some run in daily windows), and the run keeps going after you close the tab.
Credits
- Quote — every change to circuit, machine or shots re-prices through
POST /runs/quote, the same path that will charge. Cost and balance on the rail are that answer. - Hold — submitting holds the quoted amount. The dialog for hardware says it in numbers before you confirm.
- Settle — a finished run settles the hold. A free simulator run settles to zero and the credit line says "No credits were charged".
- Cancel — the Stop icon in the Runs rail cancels a queued run and returns the hold.
- Refund — a failure on the machine's side is refunded automatically; a hardware job still unfinished after 72 hours is failed and refunded. The Output pane's credit line reports exactly what happened: refunded, pending review, declined, or refund-eligible with a pointer to the console — never a rounder story than the truth.
- Not enough — the rail says Not enough credits — top up in the console and the button stays off. Top up at console.roroquantum.com.
AI credits
Copilot turns and Explain this result are metered separately as AI credits — no per-message cap; each cost is printed in the Copilot footer and on each explain card as it settles. Copilot can ask the server for a quote (a price check that holds nothing) and read this project's runs, but it has no tool that can submit a run or spend run credits — only the Run rail submits. When your balance runs out the server refuses the next request before it is charged (rounds already settled in that turn stand) and the feed shows a Not enough AI credits card with a top-up link.
Where runs live
The Runs rail lists runs submitted from this project with live status, queue position (N ahead), shots and cost; Refresh re-reads. Selecting a run opens its story in the dock. Runs belong to your account — deleting a project leaves them intact.
Related: Run a circuit · Copilot