OpenAI

Configure OpenAI as an LLM provider for AI GO! evaluations.

CLI

CLI integrations allow automating your assessment pipelines and triggering CI/CD jobs via the LatticeFlow CLI.

Step 1: Configure the integration

lf integration add --provider openai --api-key $OPENAI_API_KEY

Step 2: Use in an evaluation

Reference the model in your evaluation config.

For a list of models, please refer to the list of OpenAI models and use openai/<model_key.

models:
  - $provider: "openai/gpt-4o"

You can integrate a global provider by supplying the API key once, then creating individual models.

Step 1: Navigate to Integration Settings

Open the LatticeFlow platform and go to Settings > Integrations.

Step 2: Select OpenAI Provider

From the list of available providers, select OpenAI.

Step 3: Add your API Key

Paste your OpenAI API key into the field and save.

Step 4: Test the Integration

Run a connection test to confirm the key works.

Step 5: Navigate to AI System > Evaluate > Models

Open the AI System you are assessing, navigate to the models section, and select the model to add.

The model now appears in the list of available models to use.

You can also add individual models and customize them if needed from the AI System > Evaluate > Models page.

Step 1: Create new Custom Model Integration

From the Models page, create a new custom model integration.

Step 2: Fill in Basic Information

Assign the name and key used to refer to this model within the LatticeFlow platform.

Step 3: Fill in Connection Information

For an OpenAI model, the key fields are:

Model Adapter: select OpenAI Chat Completion
URL: the OpenAI Chat Completions endpoint, https://api.openai.com/v1/chat/completions
API Key: your OpenAI key
Custom Model Key: the OpenAI model key you would like to use

Step 4: Test the Connection

Click Create Model, select Test Connection in the top right, and click Run.