> ## Documentation Index
> Fetch the complete documentation index at: https://docs.egma.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# What are provider API keys?

Open **Settings → Provider API Keys** to manage keys for your organization.
These keys pay for calls to model providers. The separate **API keys** page
creates tokens that authenticate scripts and integrations to Egma.

An organization admin can add, replace, or remove one key for each supported
provider: OpenAI, Deepgram, and Cartesia. Other members can see which providers
have a saved key and its masked ending.

Manage these settings through a browser session or an organization API key.
A project API key cannot read or change organization provider keys, even when
its owner is an admin.

## Set up your keys

Follow [Set up provider API keys](/docs/platform/provider-api-keys/set-up-provider-api-keys) to add, replace, or remove a key.

## Inference credit on Egma Cloud

A call that uses your provider key does not deduct Egma inference credit.
The provider bills the account that owns your key directly. Usage still
appears in Egma's usage and cost records.

Funding is separate for each provider. For example, an OpenAI key pays for
OpenAI calls; Cartesia calls still use Egma credit unless you also add a
Cartesia key. Your Egma plan fee, usage allowances, and applicable voice
overages remain in effect.

Choose **Manage key → Remove key** to return a provider to the default keys.
The confirmation explains that new calls on Egma Cloud will use inference
credit again. Removing a key does not change who paid for earlier calls.

## Self-hosted deployments

Organization keys take precedence over the deployment's
[self-hosted environment settings](/self-hosting/get-started#environment-and-secrets).
