AI Provider
AI Providers are the bridge between Operaide and the large language models that power its AI features. Each provider configuration tells the platform how to connect to an external AI service — which API endpoint to call, which credentials to use, and which models to expose to reaktors and other platform features.
You manage AI Providers from Connections > AI Provider in the sidebar.
Overview
The AI Provider management page lets you:
- Register connections to one or more AI services (e.g. OpenAI, Anthropic, Azure OpenAI)
- Configure API credentials and provider-specific settings in JSON
- Define one or many model configurations per provider (each with its own model ID and parameters)
- Activate or deactivate providers without deleting them
- Test a provider's models directly from the editor
- Import and export provider configurations as JSON for backup or transfer between environments
Supported Providers
Operaide ships with built-in support for the following AI providers:
| Provider | Type identifier | Notes |
|---|---|---|
| OpenAI | openai | GPT-4o, GPT-4o-mini, o-series, and other OpenAI models |
| Anthropic | anthropic | Claude model family |
| Azure OpenAI | azure | OpenAI models hosted on Microsoft Azure |
| Azure AI | azureAi | Azure AI services beyond OpenAI |
| Google Generative | googleGenerative | Gemini models via the Google AI Studio API |
| Google Vertex | googleVertex | Gemini and other models via Google Cloud Vertex AI |
| Amazon Bedrock | amazonBedrock | Models hosted on AWS Bedrock |
| Groq | groq | High-speed inference for open-source models |
| Mistral | mistral | Mistral AI models |
| Cohere | cohere | Cohere language models |
| DeepSeek | deepseek | DeepSeek models |
| Cerebras | cerebras | Cerebras inference |
| Ollama | ollama | Self-hosted open-source models via Ollama |
Each provider type comes with an example configuration that pre-fills the JSON editor when you select it, giving you a template to adapt.
Managing AI Providers
Viewing Providers
The AI Provider list page shows all configured providers for your organisation in a table with the following columns:
| Column | Description |
|---|---|
| ID | The immutable readable ID assigned at creation |
| Name | The display name of the provider |
| Provider | The provider type (e.g. openai, anthropic) |
| Status | A badge showing Active or Inactive |
| Created At | Timestamp of when the provider was created |
| Updated At | Timestamp of the last modification |
| Actions | Edit and Delete buttons (visibility depends on your permissions) |
Use the search bar at the top to filter providers by name, readable ID, or provider type.
Creating a Provider
- Click the Create button (top-right).
- Fill in the following fields:
| Field | Description |
|---|---|
| Readable ID | A short, unique identifier with no whitespace (min. 3 characters). This cannot be changed after creation. |
| Name | A human-readable display name (min. 3 characters, must be unique within your organisation). |
| Active | Toggle to activate or deactivate the provider. Defaults to active. |
| Provider Type | Select from the list of supported providers. An icon and label help you identify each option. |
| Configuration (JSON) | A code editor where you enter the provider-specific configuration and model definitions. |
- The Configuration editor uses the following JSON structure:
{
"providerConfig": {
"apiKey": "your-api-key",
"baseURL": "https://api.example.com/v1"
},
"modelConfig": [
{
"id": "operaide.provider.model-name",
"resourceId": "actual-model-id",
"type": "chat"
}
]
}
- providerConfig contains API credentials and connection settings (differs per provider type).
- modelConfig is an array of model definitions. Each entry needs at least:
id— the application-level model ID used within Operaide (no whitespace)resourceId— the model ID as expected by the provider's APItype—"chat","embedding", or"custom"(defaults to"chat")
When you select a provider type, the editor is pre-filled with an example configuration including documentation links. Use this as a starting point and replace the placeholder values with your actual credentials and model IDs.
- Click Save to create the provider.
Editing a Provider
Open a provider by clicking Edit in the actions column. All fields are editable except the Readable ID, which is immutable after creation.
Deleting a Provider
Click Delete in the actions column. A confirmation dialog asks:
"Do you really want to delete the provider {name}?"
Confirm to permanently remove the provider and all its model configurations.
Testing a Provider
The built-in AI Provider Tester lets you verify that a provider configuration works before using it in production.
- In the provider editor, click Test Provider to open the tester modal.
- Select a Test Model from the dropdown. The list is populated from the model configurations defined in the JSON editor.
- Enter a Test Message (a default prompt is pre-filled).
- Click Send Test Message.
The tester displays results across five tabs:
| Tab | Content |
|---|---|
| Request | The raw API request object sent to the provider |
| Response | The raw API response received |
| Usage | Token counts, cost estimates, and other usage metrics |
| Warnings | Any warnings returned by the provider |
| Errors | Error details if the test failed |
The Test Response field shows the model's reply as plain text.
The tester uses the configuration as currently entered in the editor — you do not need to save the provider first. This makes it easy to experiment with different settings.
Import and Export
Exporting a Provider
In the provider editor, click Export. The full provider configuration is copied to your clipboard as a JSON object. You can paste it into a file for backup or share it with colleagues.
Importing a Provider
- On the AI Provider list page, click Import.
- A modal opens with a large text area.
- Paste a valid AI Provider JSON configuration.
- Click OK to validate and save the imported provider.
The import validates the JSON against the expected schema. If the JSON is malformed or missing required fields, you receive an error notification.
Permissions
Access to AI Provider features is controlled by the following permissions:
| Permission | Description | Default roles |
|---|---|---|
| Use AI Provider | Use AI providers for AI-powered features | Organisation Member |
| View AI Provider | View AI provider configurations | App Developer, Credential Manager |
| Edit AI Provider | Create and modify AI provider configurations | Credential Manager |
| Delete AI Provider | Delete AI provider configurations | Credential Manager |
| View Default AI Provider | View the default AI provider settings | Any logged-in user |
Only users with the Credential Manager role can create, edit, or delete AI Providers. If you need access, contact your organisation administrator.