# Usage

https://omnimodel.me/docs/usage

The console Usage page shows trends and totals, and the Billing page shows every charge. Both read from the same billing records.

## Usage summary

For the selected time range, the Usage page shows:

- Requests, successes, failures and failure rate
- Amount spent, and savings against official prices
- Tokens used and media tasks
- A daily trend, broken down by model or by key

## Billing lines

Each row on the Billing page is one charge: time, model, key prefix, quantity, unit price, amount, credit source (paid balance or gift credit), status, call ID and whether it is estimated.

A call that used both gift credit and paid balance appears as several rows; their amounts add up to the call's total. Rows marked "estimated" come from streams that were cut off before exact usage was available.

## Filtering and export

- A query can cover at most 92 days; longer ranges return `query.range_too_long`.
- Filter by model, key and status.
- CSV export produces a download link valid for 15 minutes. All times in the file are UTC.

## Time and timezone

Daily totals, today and this month follow the instance timezone; timestamps in the API and exports are UTC. Daily aggregates are recomputed every hour, so the last hour may show up in the trend a little later, while billing lines appear immediately.

## Privacy

The platform records only the usage and amounts needed for billing. Your prompts and model outputs are not stored.

## FAQ

### Why does one call show two billing lines?

It used both gift credit and paid balance, so it is split by credit source. The amounts add up to the total.

### Which timezone does the export use?

UTC. Daily totals, today and this month in the console follow the instance timezone.

### Are my prompts stored?

No. Only the usage and amounts needed for billing are recorded, not prompts or model outputs.