OpenAIProviders

Summary

The OpenAIProviders class manages AI provider configurations by loading configuration and handling resolution of model strings in the provider:model format.

For complete usage documentation, see Provider Model Aliases.

Description

This class enables multi-provider support by:

• Loading provider configurations from a single JSON file
• Loading named model aliases that map to providers and model IDs
• Resolving provider:model syntax to full API configurations
• Resolving named model aliases by bare name to full provider + model configurations

The OpenAI class automatically loads provider configurations when instantiated.

Constructor

var $providers := cs.AIKit.OpenAIProviders.new()

Creates a new instance that loads provider configuration from the AIProviders.json file (see Configuration Files in the "Provider Model Aliases" page for details on file locations and format).

Important:

• Only the first existing file is loaded. There is no merging of multiple files.
• The configuration is read once at instantiation time. If the AIProviders.json file is modified afterward, those changes will not be reflected in the existing instance. You must create a new instance of OpenAIProviders to reload the updated configuration.

Usage

Integration with OpenAI Class

var $client := cs.AIKit.OpenAI.new()

// Use model aliases with provider:model syntax
var $result := $client.chat.completions.create($messages; {model: "openai:gpt-5.1"})
var $result := $client.chat.completions.create($messages; {model: "anthropic:claude-3-opus"})
var $result := $client.chat.completions.create($messages; {model: "local:llama3"})

Direct Provider Access

var $providers := cs.AIKit.OpenAIProviders.new()

// Get a specific provider configuration
var $config := $providers.get("openai")
// Returns: {baseURL: "...", apiKey: "...", modelAliases: [...], ...} or Null

// Get all provider names
var $names := $providers.list()
// Returns: ["openai", "anthropic", "mistral", "local"]

Functions

get()

get(name : Text) : Object

Get a provider configuration by name.

Parameter	Type	Description
name	Text	The provider name
Function result	Object	Provider configuration object, or Null if not found

Example

var $config := $providers.get("openai")
If ($config # Null)
    // Use $config.baseURL, $config.apiKey, etc.

    // We could build a client with it
    var $client:=cs.AIKit.OpenAI.new($config)
End if

list()

list() : Collection

Get all provider names.

Parameter	Type	Description
Function result	Collection	Collection of provider names

Example

var $names := $providers.list()
// Returns: ["openai", "anthropic", ...]

For each ($name; $names)
    var $config := $providers.get($name)
End for each

modelAliases()

modelAliases() : Collection

Get all configured model aliases.

Parameter	Type	Description
Function result	Collection	Collection of model alias objects

Each object in the collection contains:

Property	Type	Description
name	Text	Model alias name
provider	Text	Provider name
model	Text	Model ID to use with the provider

Example

var $models := $providers.modelAliases()
// Returns: [{name: "my-gpt", provider: "openai", model: "gpt-5.1"}, ...]

For each ($model; $models)
    // $m.name, $m.provider, $m.model
End for each

Model Resolution

Two syntaxes are supported for model resolution:

Provider alias (provider:model)

Specify the provider and model name directly:

var $client := cs.AIKit.OpenAI.new()
$client.chat.completions.create($messages; {model: "openai:gpt-5.1"})

This is resolved internally to:

1. Split "openai:gpt-5.1" into provider="openai" and model="gpt-5.1"
2. Look up the "openai" provider configuration
3. Extract baseURL and apiKey
4. Make the API request using the resolved configuration

Examples:

• "openai:gpt-5.1" → Use OpenAI provider with gpt-5.1 model
• "anthropic:claude-3-opus" → Use Anthropic provider with claude-3-opus
• "local:llama3" → Use local provider with llama3 model

Model alias (bare name)

Use a named model by its bare name from the models section of the configuration:

var $client := cs.AIKit.OpenAI.new()
$client.chat.completions.create($messages; {model: ":my-gpt"})

This is resolved internally to:

1. Look up "my-gpt" in the models configuration
2. Find its provider (e.g., "openai") and model (e.g., "gpt-5.1")
3. Resolve the provider to get baseURL and apiKey
4. Make the API request using the resolved configuration

Examples:

• "my-gpt" → Use the model alias "my-gpt" (resolves to its configured provider and model)
• "my-embedding" → Use the model alias "my-embedding" for embedding operations