OpenAI Chat Feature

Feature Name	OpenAI Chat
Feature ID	CrestApps.OrchardCore.OpenAI

Provides OpenAI-compatible AI services for Orchard Core.

Overview

Use this provider for OpenAI and for OpenAI-compatible platforms such as DeepSeek, Google Gemini, Together AI, vLLM, Cloudflare Workers AI, LM Studio, LocalAI, and similar services.

appsettings.json configuration

Add an OpenAI connection under OrchardCore:CrestApps:AI:Connections:

{
  "OrchardCore": {
    "CrestApps": {
      "AI": {
        "Connections": [
          {
            "Name": "openai-cloud",
            "ClientName": "OpenAI",
            "ApiKey": "your-api-key"
          }
        ],
        "Deployments": [
          {
            "Name": "chat-default",
            "ClientName": "OpenAI",
            "ConnectionName": "openai-cloud",
            "ModelName": "gpt-4o",
            "Properties": {
              "AIDeploymentMetadata": {
                "Features": [ "textGeneration", "toolCalling", "streaming", "structuredOutputs" ]
              }
            }
          },
          {
            "Name": "embedding-default",
            "ClientName": "OpenAI",
            "ConnectionName": "openai-cloud",
            "ModelName": "text-embedding-3-large",
            "Properties": {
              "AIDeploymentMetadata": {
                "Features": [ "textEmbedding" ]
              }
            }
          },
          {
            "Name": "image-default",
            "ClientName": "OpenAI",
            "ConnectionName": "openai-cloud",
            "ModelName": "dall-e-3",
            "Properties": {
              "AIDeploymentMetadata": {
                "Features": [ "imageOutput" ]
              }
            }
          }
        ]
      }
    }
  }
}

ClientName ties a deployment to its provider and ConnectionName to the connection it authenticates with. Features declares what the model can do; which deployment serves chat, utility, embedding and the rest is decided by the deployment slots under Configuration -> Artificial Intelligence -> Settings, so one deployment can back several slots. See Model capabilities for the full list of features and slots.

Set Endpoint when the provider uses a custom OpenAI-compatible base URL:

{
  "Connections": [
    {
      "Name": "deepseek",
      "ClientName": "OpenAI",
      "Endpoint": "https://api.deepseek.com/v1",
      "ApiKey": "your-deepseek-api-key"
    }
  ]
}

Recipe setup

Use AIProviderConnections to create the connection and AIDeployment to create the deployments that profiles can select.

{
  "steps": [
    {
      "name": "AIProviderConnections",
      "Connections": [
        {
          "Source": "OpenAI",
          "Name": "deepseek",
          "DisplayText": "DeepSeek",
          "Properties": {
            "OpenAIConnectionMetadata": {
              "Endpoint": "https://api.deepseek.com/v1",
              "ApiKey": "your-deepseek-api-key"
            }
          }
        }
      ]
    },
    {
      "name": "AIDeployment",
      "Deployments": [
        {
          "Name": "deepseek-chat",
          "ModelName": "deepseek-chat",
          "ClientName": "OpenAI",
          "ConnectionName": "deepseek",
          "Properties": {
            "AIDeploymentMetadata": {
              "Features": [ "textGeneration", "toolCalling", "streaming" ]
            }
          }
        },
        {
          "Name": "deepseek-reasoner",
          "ModelName": "deepseek-reasoner",
          "ClientName": "OpenAI",
          "ConnectionName": "deepseek",
          "Properties": {
            "AIDeploymentMetadata": {
              "Features": [ "textGeneration", "reasoning", "streaming" ]
            }
          }
        }
      ]
    }
  ]
}

Selecting deployments

Choose deployments explicitly on AI profiles, templates, or chat interactions when you need provider-specific model selection. For tenant-wide fallbacks, configure Settings -> Artificial Intelligence -> Default Deployments.

Other OpenAI-compatible providers

To connect to another compatible platform, keep the same Orchard structure and change only the provider endpoint, API key, and deployment model names. For example, a Gemini connection typically uses an endpoint such as https://generativelanguage.googleapis.com/v1.