Spending
The Spending view shows where your money goes, and it keeps the two pots apart. They are different accounts in different currencies, so Daimond shows them as two tracks and never adds them together.
Two pots, kept apart
- Inference. The model calls themselves, billed to your own provider key. Daimond prices and records them on this device, per turn, so you can see the cost of your own key spend without any of it passing through us.
- Credits. The prepaid balance the gateway spends on the few things that leave the browser: fetching a page, syncing or sending mail, cross-device sync, and keeping your workspace in cloud storage.
Opening it
Open the Spending panel from the spend row at the foot of the rail, or from a link in the Credits view. It shows each pot with a headline figure, a daily graph, a breakdown by category and a plain table of movements, so a charge is never a mystery.
Protection from runaway costs
A workspace that can run many agents at once can spend fast. The danger is not a large monthly total but a rate spike: a fan-out burning through credit in seconds, before you can react. Four measures catch the spike itself, and none asks you to set a limit in advance.
- A pace check before a fan-out runs. The cost of dispatching a batch is known before any of it starts. A batch that would run faster than your own recent pace is held, and Daimond asks once, with the number and the estimate on the prompt. A few agents of ordinary cost never trip it; a sudden fifty do. Your own normal is the measure, learned from your recent spend, so there is nothing to configure.
- Pause and stop, always to hand. The Agents panel pauses every running agent at a stroke, or stops them outright, and the rail's Everything row pauses the whole app. Pausing stops the spend at once and keeps the work, so a look costs nothing.
- A separate key for every agent. Each agent running on credits spends its own freshly minted key, capped against what is left after its siblings' claims. No two agents share a key, so parallel agents cannot race a shared allowance past its cap.
- Everything metered, per turn, on your device. Every model call is priced and recorded here as it happens, feeding the graphs above. The bill arrives in front of you as it is spent, never later as a surprise.
The intent is plain: run a team of agents without fear that a runaway loop, or a fan-out larger than you realised, can cost more than a glance and a click can undo.