Budgets and receipts
See what each turn used and who paid for it, follow usage over time, and set limits for your plan, credits and your own keys.
Nib shows usage as your agents work, not at the end of the month. Every turn gets a receipt, Settings › Usage adds it up over time, and limits warn you or stop work before usage gets away from you.
Who pays, and how it shows
Each model call is paid one of four ways, and Nib shows it in that payer's own unit:
| Who pays | How it shows |
|---|---|
| Your plan (Nib models within your plan's allowance) | A share of the plan's week, or its month on plans counted by the month. Never a money amount |
| Nib credits | What was charged to your credit |
| Your organization's invoice (Team or Enterprise, beyond the plan) | What was billed beyond the plan |
| Your own keys | An estimate from the provider's published prices and the tokens it reports. Your provider bills you, and its invoice is the final word |
Models on your machine (Ollama, LM Studio, vLLM, llama.cpp) cost nothing per token. How the plan, credits and invoices work is in Plans and billing.
Receipts
When a turn finishes, the transcript shows a summary line with how long it took, how many tools ran and the tokens produced. When part of the turn was paid in money, the line adds that amount, for example $0.42 on credits or ≈ $0.31 on your Anthropic key. Work your plan covered adds nothing to the line. Click the line to open the turn's receipt.
A receipt breaks the turn down by model, including the small model Nib uses for titles and summaries, subagents, and any helper calls made during the turn:
- Input, output and cached tokens for each model.
- What each model used, in its payer's unit, and the turn's total.
- Cache saved: what prompt caching saved on this turn, shown when the whole turn was paid in money and it saved at least $0.01.
- The outcome: completed, interrupted, failed, or Paused at budget.
The Receipt pane (in the pane toggles next to Changes, Terminal and Files) lists every turn of the session, newest first. Expand a turn for its per-model split.
How amounts are shown
| You see | Meaning |
|---|---|
0.4% of this week's plan usage | Your plan covered it: its share of the plan's week (this month's on plans counted by the month). Shown in receipts and Settings › Usage, not on the turn's line |
Included in your plan | Your plan covered it, and its share isn't known yet |
$0.42 on credits | Charged to your Nib credits |
$2.10 beyond the plan | Billed to your organization beyond the plan |
≈ $0.31 on your Anthropic key | An estimate of what your provider bills you |
<$0.01 | Less than a cent |
local | A model on your machine. It costs nothing per token |
— | Nib has no price for this model. Totals leave it out, and the tooltip says so |
Without a plan, the title bar shows today's spend. On a plan, the title bar shows no spend, and Nib tells you when you've used 75%, 90% and all of the plan's week (or month).
Settings › Usage
Open Settings › Usage, or choose Usage from the composer's usage ring, the title-bar spend, the account menu, the command palette (⌘K), or the Receipt pane. Esc returns where you were. In a browser it's /account/usage.
From top to bottom:
- Limits: a card for each way of paying, then your fine rules.
- This computer: what this computer's sessions used.
- Your limits, when your organization sets them: what you've used beyond the plan, and Request more usage.
- For owners and admins: Usage by member, with the organization's activity and CSV exports.
Limits
Each card shows how much is used and what happens at its limit. A slider sets the limit and saves when you let go, and one sentence under it says the rule as it moves.
| Card | Measured in | Limit |
|---|---|---|
| Subscription | Percent of the plan's week (or month), and when it resets | None: the plan's week (or month) is the limit, and it resets on its own |
| Nib credits | Your balance, bought and gifted apart | Use credits when the plan runs out (off until you switch it on), and an optional monthly limit on what credits pay. Gifted credit is spent first. Top up and automatic top-up are underneath |
| Your keys | What your own keys spent today, this week or this month, as an estimate | A limit per day, week or month. The default is $25 a day |
Subscription shows only on a plan, and Nib credits only where your plan sells credit or you hold a balance. Your keys and the fine rules belong to this computer's engine, so they show when you're connected to it.
At the limit offers these choices:
- Warn: Nib tells you and goes on.
- Ask: Nib asks whether to go on.
- Cheaper model: Nib switches to the model you pick.
- Stop: Nib stops until the limit resets.
Nib credits offer Ask, Cheaper model and Stop: credit stops at its monthly limit, so there's no Warn. Nib notifies you at 50% and 80% of a limit.
Fine rules are limits for narrower cases, each in its own unit and saying what it counts. Add a rule asks what it applies to (a folder by day or month, all projects by day or month, or each turn), the amount, what it counts (credits and your keys, your keys only, or credits only) and what happens at the limit. A rule that counts your plan's work alongside money shows no amount. Crew members' and teams' caps are listed under Nib Crew: remove one there, or Edit in Crew to change it in the member's or team's settings. Over remote control they're read-only.
When a hard limit is reached, the session pauses on a card in the session and in your inbox: finish this turn (the limit applies again from the next one), switch to a cheaper model, stop, or, for a money limit, raise it. With At the limit set to Stop or Cheaper model, Nib does that without asking. A session no one can answer, such as a crew member's, stops.
This computer
What the sessions on this computer used, from its engine:
- Period: Today, This week (the default) or This month.
- Measure: spend or tokens. Spend is offered when something in the period was paid in money; work your plan covered is measured in tokens.
- Totals: what was used (your plan's share, and money only for what was paid in money), cache saved, tokens, and unpriced tokens (usage on models without a price).
- By day, plus tables By model, By workspace, By what ran, By identity and Most expensive sessions.
- Export CSV.
Sessions on other devices aren't included.
Limits in the Agent SDK and Nib Crew
The Agent SDK and Nib Crew have limits of their own:
Pass maxBudgetUsd (TypeScript) or max_budget_usd (Python) to a query. It caps what the query is paid in money: credits, usage billed beyond the plan, and your own keys at their estimate. Work your plan covers doesn't count toward it; the plan's own week bounds that. When the query's spend reaches the limit, Nib interrupts the run and the result has the subtype error_max_budget_usd. The result's total_cost_usd is the same figure: 0 when your plan covered the whole query.
import { query } from "@nibapp/agent-sdk";
for await (const message of query({
prompt: "Upgrade the test runner and fix what breaks",
options: { maxBudgetUsd: 2 },
})) {
if (message.type === "result") console.log(message.subtype, message.total_cost_usd);
}The check runs as usage is reported, so a run can go slightly over the limit by the cost of the model call in progress. See Agent SDK.
Next
- Plans and billing: what each plan includes, credits and invoices.
- Models and effort: pick lighter models for routine work.
- Race: compare models on the same task.
Attention inbox
One list of everything your agents are waiting on — permissions, questions, plan approvals, failing checks and pull requests — with keyboard triage.
Review
Read what the agent changed in the Changes pane, revert or accept single hunks, send comments back to the agent, and run code reviews at four depths.