# Inference Keys

Source: https://docs.sliplane.io/inference/keys

Give API keys access to Inference, cap their spend, and restrict them to confidential models.

> **info**
>
> **Private beta:** Email [support@sliplane.io](mailto\:support@sliplane.io) to request Inference access for your team.

Inference requests use regular Sliplane [API keys](/teams/api-keys) with the `inference` scope. A key can combine the `inference` scope with permissions for other Sliplane resources.

After Inference access is enabled for your team, create and manage keys under [Inference > Keys](https://sliplane.io/app/inference/keys) or on the team's [API tab](https://sliplane.io/app/team/api).

## Spend limits

A spend limit caps what a key can spend on Inference in a period. Set an amount in EUR and choose a daily, weekly, or monthly period. Periods reset at midnight UTC: daily limits every day, weekly limits on Monday, and monthly limits on the 1st.

Once a key reaches its limit, Inference requests return `429` with the message `API key spend limit reached.` until the period resets. Raising or removing the limit takes effect on the next request.

## Confidential keys

Mark a key **Confidential** for data that falls under professional secrecy according to [§ 203 StGB](https://www.gesetze-im-internet.de/stgb/__203.html). Requests with a confidential key only go to models marked **Confidential** on the [Models](https://sliplane.io/app/inference/models) page.

Confidential keys require the **Confidential Data Agreement**. The team owner signs it under [Team > Compliance](https://sliplane.io/app/team/compliance), as described for the [Confidential Region](/more/confidential-region#get-access).

A request for any other model returns `404`. `GET /v1/models` only lists the models a confidential key can use.

## Edit or delete a key

Changes to a key's spend limit or confidential setting apply to the next request. Deleted keys stop working immediately.

If a key has been exposed, delete it and create a new one.