> For the complete documentation index, see [llms.txt](https://docs.maiagent.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.maiagent.ai/tech/en/quickstart/embedding.md).
# Embedding Models
The role of Embedding models in RAG, Embedding models currently enabled in MaiAgent Cloud, the default model for on-premises deployments, and interface requirements for your own Embedding service.
## What Is Embedding?
An Embedding model converts text into a sequence of numbers (a vector). Text with similar meanings produces similar vectors, so the system can use vector distances to find semantically related passages, even when a question and a document use different wording.
In MaiAgent's RAG workflow, the Embedding model performs two tasks:
1. **Build the index**: Convert each passage from parsed knowledge base content into a vector and store it in the vector database.
2. **Convert the question**: Convert the user's question into a vector and compare its similarity with passages in the database.
RAG workflow
For the complete retrieval workflow, see [RAG Knowledge Base Retrieval System](/tech/en/quickstart/rag.md).
## What to Know Before Choosing
* **You cannot change it in the admin console after creation**: You select the Embedding model when creating a knowledge base. Vectors produced by different models cannot be compared with each other. To switch models, create a new knowledge base or contact your MaiAgent representative.
* **Image files require a multimodal model**: To add image files (such as `.png` and `.jpg`) to a knowledge base and index them, the knowledge base must use an Embedding model that supports multimodal input. Otherwise, parsing those files will fail.
* **An Agent can use multiple models**: A single Agent can connect to knowledge bases that use different Embedding models. The platform queries each knowledge base separately and combines the results.
## Models Enabled in MaiAgent Cloud
The following models are enabled in the production environment as of October 2026. The models available when creating a knowledge base are those shown in the admin console.
| Model (name in the admin console) | Provider | Supports image files |
| ---------------------------------------- | -------------------------------------------- | -------------------- |
| OpenAI text-embedding-3 | OpenAI (`text-embedding-3-large`) | No |
| Cohere Embed V4.0 | Cohere (`embed-v4.0`) | Yes |
| Cohere Embed Multilingual v3.0 (Bedrock) | AWS Bedrock (`cohere.embed-multilingual-v3`) | No |
## On-Premises Deployment
On-premises and private cloud deployments can run an Embedding model locally (the deployment package's model catalog includes `jina-embeddings-v3`, which requires a GPU), or connect to an existing model service or cloud provider. For available models and connection options, see [Connecting Model Services for On-Premises Deployments](/tech/en/platform-development/onprem-model-services.md#local-model-catalog). For Embedding support from each cloud provider, see [Cloud Model Inference API Services](/tech/en/platform-development/cloud-model-inference-api-service.md).
### Interface Requirements for Your Own Embedding Service
When connecting to a self-hosted or third-party OpenAI-compatible Embedding service, the URL must include `/v1`, and the service must accept the parameters sent by MaiAgent. Otherwise, parsing will fail for all knowledge base files. For the full requirements, see [Connecting Model Services for On-Premises Deployments: Limitations for Your Own Embedding and Reranker Services](/tech/en/platform-development/onprem-model-services.md#embedding-reranker-limits).
{% hint style="info" %}
The Embedding model determines whether relevant content can be found. If a question frequently fails to retrieve the correct passage, first check the results using [Search Test](https://docs.maiagent.ai/build/knowledge-base/test-search-result) in the admin console. Then consider adjusting the [Reranker](/tech/en/quickstart/reranker.md), the number of retrieved passages, or rebuilding the knowledge base with a different Embedding model.
{% endhint %}
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.
Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:
```
GET https://docs.maiagent.ai/tech/en/quickstart/embedding.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
.png?alt=media)