4D AIKit

Overview

4D AIKit is a built-in 4D component that enables interaction with third-party AI APIs.

OpenAI

The OpenAI class allows you to make requests to the OpenAI API.

Configuration

First of all, initialize the OpenAI client using your API key

var $client:=cs.AIKit.OpenAI.new("your api key")

For a compatible provider API, you need to configure the server URL by setting the baseURL parameter.

var $client:=cs.AIKit.OpenAI.new({apiKey: "your api key"; baseURL: "https://your.server.ai"})

or

$client.baseURL:="https://your.server.ai"

Making requests

OpenAI provides different endpoints called resources, each offering various functions.

var $result:=$client.<resource>.<function>(<parameters...>)

The $result contains the HTTPRequest, a success status, a collection of errors and more. See OpenAIResult

See some examples below.

Chat

https://platform.openai.com/docs/api-reference/chat

Completions

https://platform.openai.com/docs/api-reference/chat/create

var $messages:=[{role: "system"; content: "You are a helpful assistant."}]
$messages.push({role: "user"; content: "Could you explain me why 42 is a special number"})
var $result:=$client.chat.completions.create($messages; {model: "gpt-4o-mini"})
// result in $result.choice

Chat helper

This helper allows you to maintain a list of user messages and assistant responses.

var $helper:=$client.chat.create("You are a helpful assistant.")
var $result:=$helper.prompt("Could you explain me why 42 is a special number")
$result:=$helper.prompt("and could you decompose this number")
// conversation in $helper.messages

Vision helper

This helper enables image analysis through the chat.

var $result:=$client.chat.vision.create($imageUrl).prompt("give me a description of the image")

Images

https://platform.openai.com/docs/api-reference/images

var $images:=$client.images.generate("A futuristic city skyline at sunset"; {size: "1024x1024"}).images

Models

https://platform.openai.com/docs/api-reference/models

Get full list of models

var $models:=$client.models.list().models // you can then extract the `id`

Get one model information by id

var $model:=$client.models.retrieve("a model id").model

Files

https://platform.openai.com/docs/api-reference/files

Upload a file for use with other endpoints

var $file:=File("/path/to/your/file.jsonl")
var $result:=$client.files.create($file; "fine-tune")
var $fileId:=$result.file.id

List all files

var $files:=$client.files.list().files

Retrieve file information

var $fileInfo:=$client.files.retrieve($fileId).file

Delete a file

var $deleteResult:=$client.files.delete($fileId)

Moderations

https://platform.openai.com/docs/api-reference/moderations

var $moderation:=$client.moderations.create("This text contains inappropriate language and offensive behavior.").moderation

Asynchronous code

If you do not want to wait for the OpenAPI response when sending a request to its API, you need to use asynchronous code. The result object will be received in a callback function.

See detailed documentation for examples

License

See the LICENSE file for details

Copyright

• This library is not affiliated with, endorsed by, or officially connected to OpenAI in any way.
• "OpenAI" and any related marks are trademarks or registered trademarks of OpenAI, LLC. All rights related to OpenAI's services, APIs, and technologies remain the property of OpenAI.
• This project simply provides an interface to OpenAI’s services and does not claim any ownership over their technology, branding, or intellectual property.