Skip to main content
Add your Gravitee AI Gateway LLM Proxy endpoint to access your governed models in Arize AX’s prompt playground and LLM-as-a-judge evaluations. The LLM Proxy exposes a single OpenAI-compatible endpoint in front of OpenAI, Anthropic, Gemini, Bedrock, Vertex AI, and any OpenAI-compatible provider, so it is added as a Custom Model Endpoint — Arize AX makes use of the OpenAI Client to make calls to it.
By adding this integration, your data may be sent through your Gravitee AI Gateway to the underlying model provider for certain actions within Arize AX (e.g., prompt playground) and your account may be billed for usage.
The LLM Proxy requires a self-hosted or hybrid Gravitee APIM installation (4.10 or later) and an Enterprise license. See Proxy your LLMs for how to create the LLM Proxy API.

​
Add the Integration Using the Arize Skills

Add a Gravitee integration from your coding agent using the Arize Skills:

Create a custom model integration named 'my-gravitee' with my base URL https://my-gateway.example.com/llmtest and API key ...

​
Add the Integration from Arize AX

​
Select Custom Model Endpoints from the Providers List

The AI Provider integrations tab

The AI Provider integrations tab

​
Fill Out Your Integration Details

The AI integration details tab

The Custom Model integration configuration pane

Give your integration a name (e.g. Gravitee). Then set your base URL and credentials.
Enter the base URL as your Gateway URL plus the context path you gave the LLM Proxy API: https://<gateway-host>/<context-path>. There is no /v1 segment, and you should not include endpoint paths like /chat/completions as these are appended automatically.

​
Set Your Credentials

How you supply credentials depends on the plan type securing your LLM Proxy API:
  • API Key plan — Gravitee reads the subscription key from a custom header, X-Gravitee-Api-Key by default. Arize AX sends the API key field as an Authorization: Bearer token, so set your subscription key in the extra headers section instead: X-Gravitee-Api-Key: <subscription-api-key>.
  • JWT or OAuth2 plan — Gravitee reads a bearer token from the Authorization header, so set your token in the API key field.
  • Keyless plan — no consumer credential is required. Gravitee recommends keyless only for internal testing, since it provides no per-consumer tracking, rate limiting, or cost attribution.

​
Define Your Models

You then need to define the available model deployments. Leave OpenAI default models off and add your Gravitee model identifiers as custom model names. The LLM Proxy namespaces model IDs by context path, using a context-path:model-name format. List the exact IDs your API exposes with:

​
Advanced Settings

The AI integration details tab

The Advanced Settings configuration pane

Expand the Advanced Settings section to configure the following:
  • Supports Function Calling - Turn this on to allow the integration to use function calling features. This is on by default.
  • Authorized Orgs - Configure which organizations have access to this integration.
  • Authorized Spaces - Configure which spaces have access to this integration.

​
Add the Integration

Finally check the box to agree to the terms and conditions, then select the Add Integration button to create the integration.