Skip to main content
By default, Omni’s AI features run on LLMs that Omni hosts: Amazon Bedrock for AWS deployments and Microsoft Foundry for Azure deployments. With Bring Your Own Key (BYOK), you can also run Omni AI on your own provider key. When you use a provider on your own key, Omni sends requests for those LLMs to that provider under your credentials, not to Amazon Bedrock or Microsoft Foundry. Refer to AI data security for what data Omni shares with LLMs. You can approve more than one provider, and users can pick from LLMs on your own key and LLMs that Omni hosts at the same time. See LLM settings to learn how to approve LLMs and set the organization default.

Requirements

To use BYOK, you’ll need:
  • An Omni Enterprise plan. Contact your Omni Account Executive or Omni support to learn more.
  • Organization Admin permissions

Supported providers

Amazon Bedrock

Default provider for AWS deployments. Hosted by Omni and billed through AI credits. Listed on the Omni managed tab.

Microsoft Foundry

Default provider for Azure deployments. Hosted by Omni and billed through AI credits. Listed on the Omni managed tab.

Omni AI

Early access. Runs on an Omni-approved inference provider at no AI credit cost. Listed on the Omni managed tab.

Anthropic Direct

Direct API access to Anthropic’s Claude LLMs

Google Gemini Enterprise Agent Platform (Vertex AI)

Access to Claude LLMs through Google Vertex AI using workload identity federation (WIF)

Databricks Genie

Access to LLMs through Databricks serving endpoints

Grok (xAI)

Access to xAI’s Grok LLMs

OpenAI

Access to GPT LLMs. Also connects to OpenAI-compatible APIs, such as Azure OpenAI and Ollama, through custom model identifiers.

Snowflake Cortex

Access to Anthropic’s Claude LLMs through Snowflake Cortex