# EmbeddingProvider

The provider an embedding model is served by. Azure OpenAI is not a separate provider: use `OpenAi` with a key whose base URL is your Azure resource.

**Kind:** a string enum, used in requests and responses.

## Values

| Value | Description |
|---|---|
| `"OpenAi"` | OpenAI's embeddings API, or any OpenAI-protocol endpoint serving the same model (Azure OpenAI). |
| `"Gemini"` | Google's Gemini API (`generativelanguage.googleapis.com`). |
| `"Qwen"` | Alibaba Cloud Model Studio (DashScope), reached through its OpenAI-compatible endpoint. A Qwen key works only with its own region's workspace host, so every Qwen key needs a base URL; there is no default. |

New values can appear, so treat one you don't know as "something else", not as an error.

## Where it's used

- Inside [`EmbeddingModelDefinition`](https://docs.maitento.com/reference/types/embedding-model-definition.md)
- Inside [`PartitionEmbeddingCredential`](https://docs.maitento.com/reference/types/partition-embedding-credential.md)
- Inside [`PartitionEmbeddingProfile`](https://docs.maitento.com/reference/types/partition-embedding-profile.md)
- Inside [`PartitionEmbeddingStatus`](https://docs.maitento.com/reference/types/partition-embedding-status.md)