---
title: Configure access to LLMs
description: Elastic's AI features work with the out-of-the-box Elastic Managed LLMs or with third-party LLMs that you connect through inference endpoints or connectors...
url: https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/ai-features/llm-guides/llm-connectors
products:
  - Elastic Cloud Serverless
  - Elastic Documentation
  - Elastic Observability
  - Elastic Security
  - Elasticsearch
applies_to:
  - Elastic Cloud Serverless: Generally available
  - Elastic Stack: Generally available
---

# Configure access to LLMs
Elastic's [AI features](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/ai-features) work with the out-of-the-box Elastic Managed LLMs or with third-party LLMs that you connect through inference endpoints or connectors.

## Elastic Managed LLMs

[Elastic Managed LLMs](https://docs-v3-preview.elastic.dev/elastic/kibana/tree/main/reference/connectors-kibana/elastic-managed-llm) are available in Kibana deployments and serverless projects with an appropriate [subscription or feature tier](https://www.elastic.co/pricing). They provide immediate access to generative AI features without requiring any setup or external model integration.
Elastic Managed LLMs are available out-of-the box. They do not require manual configuration or API key management. Alternatively, you can configure and use third-party LLM connectors, such as OpenAI, Azure, or Amazon Bedrock.
To learn more about security and data privacy, refer to [Elastic Managed LLMs](https://docs-v3-preview.elastic.dev/elastic/kibana/tree/main/reference/connectors-kibana/elastic-managed-llm) which provides details on the models.
<important>
  Using Elastic Managed LLMs incurs additional costs. Refer to [Elastic Cloud pricing](https://www.elastic.co/pricing/serverless-search) for more information.
</important>


## Tested models and performance ratings

While you can connect to a wide range of LLMs, model performance varies by solution and use case. Refer to the following pages for performance information for models Elastic has tested:
- [LLM performance matrix for Observability](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/solutions/observability/ai/llm-performance-matrix)
- [LLM performance matrix for Elastic Security](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/solutions/security/ai/large-language-model-performance-matrix)
- [Recommended models for Elastic Agent Builder](/elastic/docs-content/pull/8924/explore-analyze/ai-features/agent-builder/models#recommended-models)

<note>
  Models that do not appear in these resources may still work, but Elastic hasn't tested them for those scenarios. We recommend selecting a model with strong ratings for your intended use case.
</note>


## Connect to a third-party or self-managed LLM

<important>
  The OpenAI, Amazon Bedrock, Google Gemini, and **AI Connector** connector types are being replaced by Elasticsearch inference endpoints. To configure access to an LLM provider, create an inference endpoint with the `chat_completion` task type. For steps that use your own provider account and API key, refer to [External inference](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/elastic-inference/external). To use models that Elastic hosts instead, refer to [Elastic Managed LLMs](https://docs-v3-preview.elastic.dev/elastic/kibana/tree/main/reference/connectors-kibana/elastic-managed-llm).Availability of these connector types depends on your deployment and version:
  - <applies-to>Elastic Cloud Serverless: Deprecated</applies-to> <applies-to>Elastic Stack: Planned for deprecation</applies-to> You can no longer create new connectors of these types by selecting **Create connector** in Kibana. Existing connectors and their rule actions continue to work, but the **Connectors** page hides them by default. To show them, open **Display options** and set **Deprecated** to **Show**.
  - <applies-to>Elastic Stack: Deprecated in 9.5</applies-to> These connector types are still available, but deprecated.
</important>

Follow these guides to connect to one or more third-party LLM providers:
- [Azure OpenAI](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/ai-features/llm-guides/connect-to-azure-openai)
- [Amazon Bedrock](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/ai-features/llm-guides/connect-to-amazon-bedrock)
- [OpenAI](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/ai-features/llm-guides/connect-to-openai)
- [Google Vertex](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/ai-features/llm-guides/connect-to-google-vertex)
- [Self-managed LLMs](https://docs-v3-preview.elastic.dev/elastic/docs-content/pull/8924/explore-analyze/ai-features/llm-guides/local-llms-overview)


## Preconfigured connectors

<applies-to>
  - Elastic Cloud Serverless: Unavailable
  - Elastic Stack: Generally available
</applies-to>

You can also use [preconfigured connectors](https://docs-v3-preview.elastic.dev/elastic/kibana/tree/main/reference/connectors-kibana/pre-configured-connectors) to set up third-party LLM connectors by editing the `kibana.yml` file. This lets you enable a connector for multiple spaces at once, without performing setup in the Kibana UI for each space.
<applies-to>Elastic Stack: Deprecated since 9.5</applies-to> Preconfigured Generative AI connectors are deprecated too. For new setups, use inference endpoints.
If you use a preconfigured connector for your LLM connector, we recommend adding the `exposeConfig: true` parameter to the `xpack.actions.preconfigured` section of the `kibana.yml` config file. This parameter makes debugging easier by adding configuration information to the debug logs, including which LLM the connector uses.