class OpenAI::Client
inherits Reference

Constructors

.new(auth_token : String)

View source

Methods

#chat_completion(request : ChatCompletionRequest) : ChatCompletionResponse

Given a list of messages comprising a conversation, the model will return a response.

Arguments

• request : OpenAI::ChatCompletionRequest - Request body to create a completion

Examples

client = OpenAI::Client.new ENV["OPENAI_API_KEY"]

req = OpenAI::ChatCompletionRequest.new(
  model: OpenAI::GPT3DOT5_TURBO,
  messages: [
    OpenAI::ChatCompletionMessage.new(
      role: OpenAI::ChatMessageRole::User,
      content: "Hello!"
    ),
  ]
)

puts client.chat_completion(req)

View source

#completion(request : CompletionRequest) : CompletionResponse

Given a prompt, the model will return one or more predicted completions along with the probabilities of alternative tokens at each position. Most developer should use our Chat Completions API to leverage our best and newest models.

Arguments

• request : OpenAI::CompletionRequest - Request body to create a completion

Examples

client = OpenAI::Client.new ENV["OPENAI_API_KEY"]

completion_request = OpenAI::CompletionRequest.new(
  prompt: "Say this is a test",
  model: OpenAI::GPT3DOT5_TURBO_INSTRUCT
)

puts client.completion(completion_request)

View source

#config : ClientConfig

View source

#create_assistant(request : AssistantRequest) : Assistant

Create an assistant with a model and instructions.

Arguments

• request : OpenAI::AssistantRequest - Request body to create an assistant

Examples

client = OpenAI::Client.new ENV["OPENAI_API_KEY"]

assistant = OpenAI::AssistantRequest.new(
  name: "Spanish tutor",
  instructions: "You are a personal spanish tutor, teach everyone spanish!",
  model: OpenAI::GPT4_TURBO
)

response = client.create_assistant(assistant)

View source

#create_embeddings(request : EmbeddingRequest) : EmbeddingResponse

Get a vector representation of a given input that can be easily consumed by machine learning models and algorithms.

Arguments

• request : EmbeddingRequest - Request containing text or features to embed

Examples

client = OpenAI::Client.new ENV["OPENAI_API_KEY"]

embedding_request = OpenAI::EmbeddingRequest.new(
  input: ["Your input string goes here"],
  model: OpenAI::SMALL_EMBEDDING_3,
)

response = client.create_embeddings(embedding_request)

View source

#list_assistants(limit : Int32 | Nil = nil, order : String | Nil = nil, after : String | Nil = nil, before : String | Nil = nil) : AssistantsList

List assistants

Arguments

• limit : Int32 | Nil - How many assistants to return
• order : String | Nil - asc or desc
• after : String | Nil - assistant_id to return after
• before : String | Nil - assistant_id to return before

Examples

client = OpenAI::Client.new ENV["OPENAI_API_KEY"]

response = client.retrieve_assistant(assistant_id)

View source

#modify_assistant(assistant_id : String, request : AssistantRequest) : Assistant

Modify an existing assistant

Arguments

• assistant_id : String - Existing assistant id
• request : OpenAI::AssistantRequest - Request body to update existing assistant

Examples

client = OpenAI::Client.new ENV["OPENAI_API_KEY"]

assistant = OpenAI::AssistantRequest.new(
  name: "Spanish tutor",
  instructions: "You are a personal spanish tutor, teach everyone spanish!",
  model: OpenAI::GPT4_TURBO
)

response = client.modify_assistant(assistant_id, assistant)

View source

#retrieve_assistant(assistant_id : String) : Assistant

Retrieve an already created assistant

Arguments

• assistant_id : String - ID of created assistant

Examples

client = OpenAI::Client.new ENV["OPENAI_API_KEY"]

response = client.retrieve_assistant(assistant_id)

View source