OpenAI

This community-maintained library, written in Swift, provides an easy way to use the OpenAI REST API.

Installation

Swift Package Manager

dependencies: [
    .package(url: "https://github.com/exyte/OpenAI")
]

CocoaPods

To install OpenAI, simply add the following line to your Podfile:

pod 'ExyteOpenAI'

Carthage

To integrate OpenAI into your Xcode project using Carthage, specify it in your Cartfile

github "exyte/OpenAI"

Requirements

• iOS 16+, tvOS 16+, macOS 13+, watchOS 8+
• Xcode 15+

Development Roadmap

• [x] Models
• [x] Files
• [x] Assistants
• [x] Run streaming
• [x] Chat
• [x] Swift Concurrency support
• [ ] Audio
• [ ] Images
• [ ] Moderations
• [ ] Fine-tuning
• [ ] Vector Store Files

Assistants API usage

1. Obtain your API key. Do not share this with others or expose it in any client-side code.

   ⚠️ OpenAI strongly recommends developers of client-side applications proxy requests through a separate backend service to keep their API key safe. API keys can access and manipulate customer billing, usage, and organizational data, so it's a significant risk to expose them.

2. Create a client instance.

   let client = OpenAI(apiKey: "YOUR_API_KEY_HERE")

3. Create an Assistant by defining its instructions and model.

   let assistantPayload = CreateAssistantPayload(model: .gpt_4o, name: "My Assistant", instructions: "Be funny")
   client.createAssistant(from: assistantPayload) <...>

4. Create a Thread to start the conversation.

   let threadPayload = CreateThreadPayload(messages: [...], metadata: [...])
   client.createThread(from: threadPayload) <...>

5. Add Messages to the Thread from the user.

   let messagePayload = CreateMessagePayload(role: .user, content: "Hello!")
   client.createMessage(in: threadId, payload: messagePayload) <...>

6. Run the Assistant on the Thread to generate a response.

   client.createRun(in: threadId, payload: CreateRunPayload(assistantId: assistantId)) <...>

7. Check the Run status until it is completed or failed.

   client.retrieveRun(id: runId, from: threadId)

8. Retrieve the Messages from the Assistant.

   let listPayload = ListPayload(after: lastMessageId)
   client.listMessages(from: threadId, payload: listPayload) <...>

For more detailed information about OpenAI Assistants API usage, please refer to platform.openai.com and our Examples section.

Available endpoints

Chat

Creates a model response for the given chat conversation.

Create chat completion

createChatCompletion(
    from: CreateChatCompletionPayload(
        model: model,
        messages: [ChatCompletionMessage(role: .user, content: "Hello")]
    )
)

Returns ChatCompletion

Models

List and describe the various models available in the API. A list of models and their differences can be found on platform.openai.com.

List models

listModels()

Returns ObjectList<Model>

Retreive model

retrieveModel(with: modelId)

Returns Model

Delete a fine-tuned model

deleteModel(with: modelId)

Returns DeletionStatus

Files

Files are used to upload documents that can be used with features like Assistants.

Upload file

uploadFile(payload: FilePayload(purpose: filePurpose, fileURL: fileURL))

Returns File

List files

listFiles()

Returns ObjectList<File>

Retreive file

retrieveFile(id: fileId)

Returns File

Delete file

deleteFile(id: fileId)

Returns DeletionStatus

Retrieve file content

retrieveFileContent(id: fileId, destinationURL: destinationURL)

Returns URL

Assistants

Build assistants that can call models and use tools to perform tasks.

Create assistant

createAssistant(from: CreateAssistantPayload(model: model, name: name, ...))

Returns Assistant

List assistants

listAssistants(payload: ListPayload(limit: limit, ...))

Returns ObjectList<Assistant>

Retrieve assistant

retrieveAssistant(id: assistantId)

Returns Assistant

Modify assistant

modifyAssistant(id: assistandId, payload: CreateAssistantPayload(model: updatedModel, name: updatedName, ...))

Returns Assistant

Delete assistant

deleteAssistant(id: assistantId)

Returns DeletionStatus

Create thread

createThread(
    from: CreateThreadPayload(
    messages: [CreateMessagePayload(role: .user, content: "Hello"), ...],
    metadata: ["key1": "value1", ...]
    )
)

Returns Thread

Retrieve thread

retrieveThread(id: threadId)

Returns Thread

Modify thread

modifyThread(id: threadId, payload: ModifyPayload(metadata: ["key1": "value1", ...]))

Returns Thread

Delete thread

deleteThread(id: threadId)

Returns DeletionStatus

Create message

createMessage(in: threadId, payload: CreateMessagePayload(role: .user, content: "Hello"))

Returns Message

List messages

listMessages(from: threadId, payload: ListPayload(limit: limit))

Returns ObjectList<Message>

Retrieve message

retrieveMessage(id: messageId, from: threadId)

Returns Message

Modify message

modifyMessage(id: messageId, from: threadId, payload: ModifyPayload(metadata: ["key1": "value1", ...]))

Returns Message

Create run

createRun(in: threadId, payload: CreateRunPayload(assistantId: assistantId, ...))

Returns Run

Create run with streaming

createStreamRun(in: threadId, payload: CreateStreamRunPayload(assistantId: assistantId))

Returns StreamEvent sequence

Create thread and run

createThreadAndRun(
    from: CreateThreadAndRunPayload(
    assistantId: assistantId,
    thread: CreateThreadPayload(
        messages: [CreateMessagePayload(role: .user, content: "Hello"), ...],
        metadata: ["key1": "value1", ...]
    )
    )
)

Returns Run

List runs

listRuns(from: threadId, payload: ListPayload(limit: limit, ...))

Returns ObjectList<Run>

Retrieve run

retrieveRun(id: runId, from: threadId)

Returns Run

Modify run

modifyRun(id: runId, from: threadId, payload: ModifyPayload(metadata: ["key1": "value1", ...]))

Returns Run

Cancel run

cancelRun(id: runId, from: threadId)

Returns Run

List run steps

listRunSteps(from: runId, in: threadId, payload: ListPayload(limit: limit, ...))

Returns ObjectList<RunStep>

Retrieve run step

retrieveRunStep(id: runStepId, from: runId, in: threadId)

Returns RunStep

Examples

To try the OpenAIAssistants examples:

• Clone the repo https://github.com/exyte/OpenAI
• Open OpenAIAssistantsExample/OpenAIAssistantsExample.xcodeproj
• Try it!

Our other open source SwiftUI libraries

PopupView - Toasts and popups library
Grid - The most powerful Grid container
ScalingHeaderScrollView - A scroll view with a sticky header which shrinks as you scroll
AnimatedTabBar - A tabbar with a number of preset animations
MediaPicker - Customizable media picker
Chat - Chat UI framework with fully customizable message cells, input view, and a built-in media picker
AnimatedGradient - Animated linear gradient
ConcentricOnboarding - Animated onboarding flow
FloatingButton - Floating button menu
ActivityIndicatorView - A number of animated loading indicators
ProgressIndicatorView - A number of animated progress indicators
FlagAndCountryCode - Phone codes and flags for every country
SVGView - SVG parser
LiquidSwipe - Liquid navigation animation