External Model Connection APIs¶

Use the Connections API to programmatically manage external model connections for OpenAI, Azure OpenAI, Anthropic, custom API providers, and other integrations.

Prerequisites

Before using these APIs, ensure your API App has the following scopes enabled:

View Connections – Required for viewing connection details
Manage Connections – Required for creating and updating connections

Base URL¶

https://{host}/api/public

Common Request Headers¶

Header	Required	Description
`x-api-key`	Yes	API key for authentication
`Content-Type`	Yes	`application/json`

API List¶

API	Description	Method	Endpoint
List All Connections	Retrieves all configured connections	GET	`/connections`
Get Connection by ID	Retrieves a specific connection	GET	`/connections/{connectionId}`
Create Connection	Creates a new connection	POST	`/connections`
Update Connection	Updates an existing connection	PATCH	`/connections/{connectionId}`

List All Connections¶

Retrieves a list of all configured connections with optional filtering.

Method	Endpoint
GET	`https://{{host}}/api/public/connections`

Query Parameters¶

Parameter	Required	Type	Description
`provider`	No	String	Filter by provider. Accepted values: `Open AI`, `Anthropic`, `Azure Open AI`, `API`
`status`	No	String	Filter by status. Accepted values: `ACTIVE`, `INACTIVE`
`limit`	No	Integer	Maximum number of results to return

Sample Request¶

curl --location 'https://{{host}}/api/public/connections?limit=10' \
--header 'x-api-key: {{apiKey}}` \
--header 'Content-Type: application/json'

Sample Response¶

{

  "connections": [
    {
      "connectionId": "cn-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
      "connectionName": "OpenAI GPT-4o",
      "provider": "Open AI",
      "model": "gpt-4o",
      "status": "ACTIVE",
      "createdOn": "2024-01-15T10:30:00.000Z",
      "modifiedOn": "2024-01-15T10:30:00.000Z"
    }
  ],
  "total": 1
}

Get Connection by ID¶

Retrieves the details of a specific connection using its unique identifier.

Method	Endpoint
GET	`https://{{host}}/api/public/connections/{connectionId}`

Path Parameters¶

Parameter	Required	Type	Description
`connectionId`	Yes	String	The unique identifier of the connection

Sample Request¶

curl --location 'https://{{host}}/api/public/connections/{{connectionId}}` \
--header 'x-api-key: {{apiKey}}' \
--header 'Content-Type: application/json'

Sample Response¶

{
  "connectionId": "cn-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
  "connectionName": "OpenAI GPT-4o",
  "provider": "Open AI",
  "model": "gpt-4o",
  "status": "ACTIVE",
  "fields": {
    "API_KEY": "sk-***************"
  },
  "createdOn": "2024-01-15T10:30:00.000Z",
  "modifiedOn": "2024-01-15T10:30:00.000Z"
}

Create Connection¶

Creates a new connection to an external model provider. Supports creation of Custom API, OpenAI, Azure OpenAI, Anthropic, and other external model connections.

Method	Endpoint
POST	`https://{{host}}/api/public/connections`

Body Parameters

Parameter	Required	Type	Description
`provider`	Yes	String	The provider type. Accepted values: `API`, `Open AI`, `Azure Open AI`, or `Anthropic`
`connectionName`	Yes	String	Display name for the connection
`model`	Conditional	String	Model identifier. Required for `Open AI`, `Azure Open AI`, and `Anthropic` providers
`modelType`	Conditional	String	Set to `EASY_INTEGRATION` for `Azure Open AI` and `Anthropic` providers
`fields`	Yes	Object	Provider-specific configuration fields

Sample Request – Custom API Connection - Default Provider¶

curl --location '{{BASE_URL}}/api/public/connections' \
--header 'x-api-key: {{API_KEY}}' \
--header 'Content-Type: application/json' \
--data '{
  "model": "CustomProvider",
  "provider": "API",
  "connectionName": "CustomProvider",
  "promptVars": [
    {
      "name": "prompt",
      "displayName": "Prompt",
      "status": true,
      "dataType": "String",
      "elementType": "textBox",
      "defaultValue": "hlo",
      "required": true
    },
    {
      "name": "system_prompt",
      "displayName": "System prompt",
      "status": false,
      "dataType": "String",
      "elementType": "textBox",
      "defaultValue": "",
      "required": false
    },
    {
      "name": "examples",
      "displayName": "Examples",
      "status": false,
      "dataType": "String",
      "elementType": "textBox",
      "examples": true,
      "defaultValue": "",
      "required": false
    }
  ],
  "customVars": [],
  "endpointUrl": "{{LLM_API_URL}}",
  "headers": [
    {
      "key": "x-api-key",
      "value": "{{LLM_API_KEY}}"
    }
  ],
  "payload": {
    "model": "{{MODEL_NAME}}",
    "max_tokens": 1024,
    "messages": [
      {
        "role": "user",
        "content": "What is the weather like in San Francisco?"
      }
    ]
  },
  "outputPath": "content[0].text",
  "inputTokensPath": "usage.input_tokens",
  "outputTokensPath": "usage.output_tokens",
  "status": "FINALIZED",
  "isEnable": true
}'

Custom API Fields (provider: "API")

Field	Required	Type	Description
`fields.BASE_URL`	Yes	String	Base URL of the API endpoint
`fields.API_KEY`	Yes	String	API key for authentication
`fields.MODEL_NAME`	Yes	String	Model name
`LLM_API_URL`	Yes	String	URL for the model endpoint
`LLM_API_KEY`	Yes	String	API key of the model

Sample Request – Custom API Connection - Existing Provider¶

curl --location '{{BASE_URL}}/api/public/connections' \
--header 'x-api-key: {{API_KEY}}' \
--header 'Content-Type: application/json' \
--data '{
  "model": "{{MODEL_NAME}}",
  "provider": "API",
  "connectionName": "Custom_Existingnew",
  "endpointUrl": "{{LLM_API_URL}}",
  "headers": [
    {
      "key": "x-api-key",
      "value": "{{LLM_API_KEY}}"
    }
  ],
  "status": "FINALIZED",
  "isEnable": true,
  "idp": "none",
  "bodyTab": "providerReferences",
  "mapProvider": "anthropicModel",
  "llmFeatures": {
    "toolCalling": false,
    "supportTools": true,
    "parallelToolCalling": true,
    "structuredResponse": true,
    "dataGeneration": false,
    "streaming": true
  },
  "IOMappings": [
    "textToText",
    "textToImage"
  ]
}'

Custom API Fields (provider: "API")

Field	Required	Type	Description
`fields.BASE_URL`	Yes	String	Base URL of the API endpoint
`fields.API_KEY`	Yes	String	API key for authentication
`fields.MODEL_NAME`	Yes	String	Model name
`LLM_API_URL`	Yes	String	URL for the model endpoint
`LLM_API_KEY`	Yes	String	API key of the model
`fields.IOMapping`	Yes	String	Supported values: "textToText", "textToImage", "imageToText", and "audioToText"
`fields.mapProvider`	Yes	String	Supported values: "anthropicModel", "geminiModel", and "openAIModel"

Sample Request – OpenAI Connection¶

curl --location 'https://{{host}}/api/public/connections' \
--header 'x-api-key: {{apiKey}}' \
--header 'Content-Type: application/json' \
--data '{
  "provider": "Open AI",
  "connectionName": "OpenAI GPT-4o",
  "model": "gpt-4o",
  "fields": {
    "API_KEY": "sk-***************"
  }
}'

OpenAI Field (provider: "Open AI")

Field	Required	Type	Description
`fields.API_KEY`	Yes	String	Your OpenAI API key

Sample Request – Azure OpenAI Connection¶

curl --location 'https://{{host}}/api/public/connections' \
--header 'x-api-key: {{apiKey}}' \
--header 'Content-Type: application/json' \
--data '{
  "connectionName": "Azure GPT-4",
  "provider": "Azure Open AI",
  "model": "GPT-4",
  "fields": {
    "API_KEY": "***************",
    "api_version": "2024-08-01-preview",
    "your_resource_name": "your-azure-resource",
    "deployment_id": "your-deployment-id"
  },
  "modelType": "EASY_INTEGRATION"
}'

Azure OpenAI Fields (provider: "Azure Open AI")

Field	Required	Type	Description
`fields.API_KEY`	Yes	String	Your Azure OpenAI API key
`fields.api_version`	Yes	String	API version (for example, `2024-08-01-preview`)
`fields.your_resource_name`	Yes	String	Your Azure resource name
`fields.deployment_id`	Yes	String	Your deployment ID

Sample Request – Anthropic Connection¶

curl --location 'https://{{host}}/api/public/connections' \
--header 'x-api-key: {{apiKey}}' \
--header 'Content-Type: application/json' \
--data '{
  "connectionName": "Anthropic Claude",
  "provider": "Anthropic",
  "model": "claude-3-5-haiku-20241022",
  "fields": {
    "API_KEY": "***************"
  },
  "modelType": "EASY_INTEGRATION"
}'

Anthropic Field (provider: "Anthropic")

Field	Required	Type	Description
`fields.API_KEY`	Yes	String	Your Anthropic API key

Sample Response¶

{
  "connectionId": "cn-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
"fields": {
       "API_KEY": "sk***************MQAA"
   },
  "connectionName": "OpenAI GPT-4o",
  "provider": "Open AI",
  "model": "gpt-4o",
  "status": "ACTIVE",
  "createdOn": "2024-01-15T10:30:00.000Z"
}

Update Connection¶

Updates the configuration of an existing connection. Supported only for external provider connections where provider is not API.

Method	Endpoint
PATCH	`https://{{host}}/api/public/connections/{connectionId}`

Path Parameters

Parameter	Required	Type	Description
`connectionId`	Yes	String	The unique identifier of the connection

Sample Request¶

curl --location --request PATCH 'https://{{host}}/api/public/connections/{{connectionId}}' \
--header 'x-api-key: {{apiKey}}' \
--header 'Content-Type: application/json' \
--data '{
  "fields": {
    "API_KEY": "sk-***************************"
  }
}'

Body Parameters

Parameter	Required	Type	Description
`fields`	Yes	Object	Updated configuration field, API_KEY

Sample Response¶

{
  "connectionId": "cn-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx", 
"fields": {
       "API_KEY": "sk*****************MQAA"
   },
  "connectionName": "OpenAI GPT-4o",
  "provider": "Open AI",
  "model": "gpt-4o",
  "status": "ACTIVE",
  "modifiedOn": "2024-01-16T14:45:00.000Z"
}

Custom API Connection¶

Custom API connections let you integrate external models by configuring your own API endpoint. You control the endpoint URL, authentication, request payload, and response mappings. This functionality applies only when the provider type is set to API.

Connection Types¶

Custom API connections come in two types:

Default Provider

You define the complete request structure from scratch.
Full control over payload format, variable placeholders, and response parsing.
Use when integrating with a custom or proprietary API that doesn't follow standard LLM provider formats.

Existing Provider

Uses an existing provider's request/response format such as OpenAI, Anthropic, Cohere, or Gemini.
You only specify the endpoint and which provider format to use.
Use when integrating with OpenAI-compatible APIs (like Groq, Together AI) or other providers that follow standard formats.

Endpoint

Method	Endpoint	Update Type	Description
`PATCH`	`https://{{host}}/api/public/connections/{connectionId}`	Partial	Modifies only specified fields; preserves everything else.
`PUT`	`https://{{host}}/api/public/connections/{connectionId}`	Full	Replaces the entire config. Requires all mandatory fields.

Path Parameters

Parameter	Required	Type	Description
`connectionId`	Yes	String	The unique identifier of the Custom API connection

Partial Update (PATCH)¶

Use PATCH for quick changes, such as rotating an API key or updating a model name, without sending the entire configuration object. Only the fields you include in the request will be updated.

Sample Request: Default Provider (Updating Endpoint and Headers)¶

This example shows how to update connections where you define the complete request structure.

curl --request PATCH 'https://{{host}}/api/public/connections/{connectionId}' \
  --header 'x-api-key: {{apiKey}}' \
  --header 'Content-Type: application/json' \
  --data '{
    "connectionName": "Updated Custom Provider",
    "endpointUrl": "https://api.example.com/v2",
    "headers": [{"key": "x-api-key", "value": "new-key-123"}],
    "outputPath": "result.text"
  }'

Body Parameters

Parameter	Type	Description
`connectionName`	String	Display name for the connection
`model`	String	Model identifier
`endpointUrl`	String	API endpoint URL
`headers`	Array	HTTP headers to include in requests
`status`	String	Connection status: `DRAFT` or `FINALIZED`

API Fields

Field	Type	Description
`promptVars`	array	Variable definitions for prompts
`customVars`	array	Custom variable definitions
`payload`	object	Request body template with variable placeholders
`outputPath`	string	JSONPath to extract response text
`inputTokensPath`	string	JSONPath to extract input token count
`outputTokensPath`	string	JSONPath to extract output token count

Sample Request: Existing Provider (Updating Features)¶

This example shows how to update connections that use standard provider formats (OpenAI, Anthropic, etc.).

curl --request PATCH 'https://{{host}}/api/public/connections/{connectionId}' \
  --header 'x-api-key: {{apiKey}}' \
  --header 'Content-Type: application/json' \
  --data '{
    "connectionName": "Groq LLaMA Updated",
    "model": "llama3-8b-8192",
    "llmFeatures": {
      "toolCalling": true,
      "streaming": true,
      "supportTools": true,
      "parallelToolCalling": false,
      "structuredResponse": false,
      "dataGeneration": false
    }
  }'

Body Parameters

Parameter	Type	Description
`connectionName`	String	Display name for the connection
`model`	String	Model identifier
`endpointUrl`	String	API endpoint URL
`headers`	Array	HTTP headers to include in requests
`status`	String	Connection status: `DRAFT` or `FINALIZED`

API Fields

Field	Type	Description
`mapProvider`	String	Tells the system which template to use: `openAIModel` or `anthropicModel`.
`llmFeatures`	Object	A set of boolean flags (e.g., `streaming: true`, `toolCalling: true`).
`IOMappings`	Array	List of supported modalities (e.g., `["textToText", "imageToText"]`).
`idp`	string	Identity provider setting; usually set to `"none"`.

Full Replace (PUT)¶

Use PUT when you need to completely reconfigure a connection or when moving a connection status from DRAFT to FINALIZED. You must provide all mandatory fields for the connection type.

Sample Request: Default Provider (Full Payload)¶

This example shows how to use PUT for a Default Provider connection. You must include the complete configuration, including all prompt variables, payload structure, and output paths.

curl --request PUT 'https://{{host}}/api/public/connections/{connectionId}' \
  --header 'x-api-key: {{apiKey}}' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "CustomProvider-v2",
    "connectionName": "Fully Updated Custom Provider",
    "endpointUrl": "https://api.newprovider.com/v1/completions",
    "headers": [
      {"key": "Authorization", "value": "Bearer new-secret-key"},
      {"key": "Content-Type", "value": "application/json"}
    ],
    "promptVars": [
      {"name": "prompt", "displayName": "Prompt", "status": true, "dataType": "String", "elementType": "textBox", "defaultValue": "", "required": true},
      {"name": "system_prompt", "displayName": "System Prompt", "status": true, "dataType": "String", "elementType": "textBox", "defaultValue": "You are a helpful assistant", "required": false},
      {"name": "examples", "displayName": "Examples", "status": false, "dataType": "String", "elementType": "textBox", "examples": true, "defaultValue": "", "required": false}
    ],
    "customVars": [],
    "payload": {
      "model": "gpt-4-turbo",
      "max_tokens": 4096,
      "temperature": 0.8,
      "messages": [
        {"role": "system", "content": "{{system_prompt}}"},
        {"role": "user", "content": "{{prompt}}"}
      ]
    },
    "outputPath": "choices[0].message.content",
    "inputTokensPath": "usage.prompt_tokens",
    "outputTokensPath": "usage.completion_tokens",
    "status": "FINALIZED"
  }'

Body Parameters

Parameter	Type	Description
`connectionName`	String	Display name for the connection
`model`	String	Model identifier
`endpointUrl`	String	API endpoint URL
`headers`	Array	HTTP headers to include in requests
`status`	String	Connection status: `DRAFT` or `FINALIZED`

API Fields

Field	Type	Description
`promptVars`	array	Variable definitions for prompts
`customVars`	array	Custom variable definitions
`payload`	object	Request body template with variable placeholders
`outputPath`	string	JSONPath to extract response text
`inputTokensPath`	string	JSONPath to extract input token count
`outputTokensPath`	string	JSONPath to extract output token count

Sample Request: Existing Provider (Full Payload)¶

This example shows how to use PUT for an Existing Provider connection. You must specify the provider format to use and the LLM capabilities.

curl --request PUT 'https://{{host}}/api/public/connections/{connectionId}' \
  --header 'x-api-key: {{apiKey}}' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "llama3-70b-8192-updated",
    "connectionName": "Groq LLaMA Full Update",
    "endpointUrl": "https://api.groq.com/openai/v1/chat/completions",
    "headers": [
      {"key": "Authorization", "value": "Bearer gsk_newkey12345"},
      {"key": "Content-Type", "value": "application/json"}
    ],
    "mapProvider": "openAIModel",
    "llmFeatures": {
      "toolCalling": true,
      "supportTools": true,
      "parallelToolCalling": true,
      "structuredResponse": true,
      "dataGeneration": false,
      "streaming": true
    },
    "IOMappings": ["textToText", "imageToText"],
    "status": "FINALIZED",
    "idp": "none"
  }'

Body Parameters

Parameter	Type	Description
`connectionName`	String	Display name for the connection
`model`	String	Model identifier
`endpointUrl`	String	API endpoint URL
`headers`	Array	HTTP headers to include in requests
`status`	String	Connection status: `DRAFT` or `FINALIZED`

API Fields

Field	Type	Description
`mapProvider`	String	Tells the system which template to use: `openAIModel` or `anthropicModel`.
`llmFeatures`	Object	A set of boolean flags (for example: `streaming: true`, `toolCalling: true`).
`IOMappings`	Array	List of supported modalities (for example: `["textToText", "imageToText"]`).
`idp`	string	Identity provider setting; usually set to `"none"`.

Send Feedback