Azure OpenAI AI Model Configuration in ConvyAI
Overview
The Azure OpenAI integration lets ConvyAI use OpenAI models deployed within your Azure subscription. This is the recommended option if your organization requires data to remain within Azure's cloud infrastructure.
Prerequisites
- An Azure OpenAI resource with a deployed model.
- The resource endpoint URL and API key.
- Credentials configured in Generative AI Settings.
Create an Azure OpenAI AI Model
Navigate to Generative AI → AI Models and click New Model → Azure OpenAI.
Configuration options
| Field | Description |
|---|---|
| Name | A unique name to identify this AI Model. |
| Key | The API key associated with your Azure OpenAI resource. |
| Endpoint | The HTTPS endpoint of your Azure OpenAI resource (e.g., https://<your-resource>.openai.azure.com/). |
| Intent threshold | Minimum confidence score for an intent to be considered valid. |
Advanced parameters
| Parameter | Default | Description |
|---|---|---|
| Prompt version | v3 | System prompt version applied when using the model in Knowledge Base nodes. |
| Completion model | gpt-5.1 | The deployment name of the model to use. This is the name you gave the deployment in Azure, not the underlying model name. |
| Embedding model | text-embedding-3-small | The model used for embedding during Knowledge Base indexing and retrieval. |
| Max token completion | 1000 | Maximum number of tokens in the generated response. |
| Temperature | - | Controls response creativity. Lower values (0.0–0.3) produce more deterministic outputs; higher values (0.7–1.0) produce more varied responses. |
| Top P | - | Nucleus sampling parameter. Limits token selection to the top probability mass. |
| Frequency penalty | - | Reduces repetition of tokens that have already appeared in the output. |
| Presence penalty | - | Encourages the model to introduce new topics by penalizing tokens that have already appeared. |
| Reasoning effort | - | The reasoning level of the model, as a free-text field that is not validated. Enter the value expected by your model (for example low, medium or high), or the value that turns reasoning off (for example none). The value applies to both Knowledge Base and AI Agent nodes, so you configure it once. |
| Number of extracted chunks | 3 | How many Knowledge Base chunks are retrieved and passed to the model as context during RAG. |
Optional parameters start empty on new AI Models. An empty field means the parameter is not sent and the provider default applies; any value you enter is sent as is. See Optional parameters.
Intents, entities, and testing
Configuring intents and entities and testing the model are features shared across all AI Model providers. See Intents, Entities & Testing for full documentation.