# Pricing and limits

> Pay for input tokens. Plan around the limits of each request and account.

Canonical: https://platform.uprelic.com/docs/pricing

---

::: definitions
## Price

Per 1M input tokens. Output tokens are free. The price is the same on every plan.

- `strom-fast` (`strom-1.0.7`): $0.042
- `strom-base` (`strom-base-1.0`, beta): $0.24

## Plans

Every account starts on Free. Paid plans add a monthly fee, plus VAT, and you pay for usage on top at the prices on this page. Change your plan under [Billing](https://platform.uprelic.com/console/billing).

| | Free | Starter | Growth | Enterprise |
|---|---|---|---|---|
| Fee a month | 0 | €135 | €485 | Contact us |
| Fee a year | 0 | €1,345 | €4,830 | Contact us |
| Seats | 2 | Unlimited | Unlimited | Unlimited |
| Request contents kept | 7 days | 30 days | 180 days | 365 days |
| Zero data retention | $50, €50, £50 or CHF 50 a month | Included | Included | Included |
| Agent guard sessions a month | None | 50,000 | 500,000 | By contract |
| Self-improvement of document extraction | No | Yes | Yes | Yes |

The fees are the same number in US dollars ($135 and $485), and £115 and £415, or CHF 125 and CHF 450. Growth starts with a 14-day free trial. On paid plans, your requests are served before Free accounts' requests when Strom is busy. Switching between paid plans applies at once: the rest of the billing period is charged or credited. Moving to Free takes effect at the end of the period you paid for.

## Spend cap

Set a monthly spend cap under Billing. A request that would take this month's spend past the cap gets `402` until the month ends or you raise the cap. Owners get an email at 80%.

## Zero data retention

20% more per 1M input tokens: $0.0504 for `strom-fast`, $0.288 for `strom-base`. Output tokens are free. On a paid plan, private requests are paid from your balance. On Free, they are paid from the ZDR credit of the [zero data retention subscription](https://platform.uprelic.com/docs/zero-data-retention): $50, €50, £50 or CHF 50 a month, plus VAT, and from ZDR top-ups, which do not expire.

## Agent guard

Input tokens at the `strom-fast` price, plus $0.50 per 1,000 sessions past your plan's monthly sessions. A check that no rule applies to is free. See [Agent guard](https://platform.uprelic.com/docs/agent-guard#cost).

## Document extraction

Input tokens at the model's price. See [Document extraction](https://platform.uprelic.com/docs/document-extraction).

## Self-improvement

Each run that improves your document extraction is charged to your balance for the AI processing it uses, at twice our provider's price. A typical run costs about €2 to €3.

## Input tokens

Calculated as: state + question list + image tokens.

Images: one token per 1,024 pixels. Images over 1 megapixel are scaled down to 1 megapixel and small images are scaled up, so an image costs between 64 and 1,024 tokens.

A [box question](https://platform.uprelic.com/docs/concepts#box) (beta) costs as much as one question plus the tokens of a copy of its image for each region it considers, at most 8. With `multiple` set to `true`, it costs one question per region instead of one in all. The images themselves are paid for once, only when other questions see them, or when a box question searches another image of the request.

## Request size

Up to 256 questions, 256 options per question, 8 images of 10 MB each, and 32,000 input tokens in all.

## Rate limits

Up to 512 questions in progress at once per account on `strom-fast` (64 on `strom-base`), counted over all your requests. When Strom is busy with all customers’ requests, new ones get `429` with a `Retry-After` header.

Contact [sales@uprelic.com](mailto:sales@uprelic.com) for higher limits.
:::
