Create Chat Completion

Given a list of messages comprising a conversation, the provided model returns an appropriate response.

For more details, see Create chat documentation in the OpenAI API reference and the related Chat Completions API guide.

Input

Field	Definition	Type	Required
Model	The identifier of the model to use. See the Model endpoint compatibility table for details on which models work with the Chat API.	Dropdown	TRUE
Messages	A list of the messages that comprise the conversation.	List of Objects	TRUE
Temperature	The sampling temperature to use in the output. Values can be from `0` through `2`. The default is `1`. Higher values like `0.8` make the output more random, while lower values like `0.2` make it more focused and deterministic. Okta generally recommends altering this input or Top P, but not both.	Number	FALSE
Top P	An alternative to sampling with temperature, this is called nucleus sampling. The model considers the results of the tokens with Top P probability mass. A value of `0.1` means only the tokens comprising the top 10% of the probability mass are considered. The default is `1`. Okta generally recommends altering this input or Temperature but not both.	Number	FALSE
N	How many chat completion choices to generate for each input message. The default is `1`.	Number	FALSE
Stop	You can specify up to four text sequences that tell the API to stop generating further tokens. See How do I use Stop Sequences? The returned chat completion doesn't contain the `stop` sequence.	Text	FALSE
Max Tokens	The maximum number of tokens to generate in the chat completion. See Tokenizer.	Number	FALSE
Presence Penalty	A number from `-2.0` through `2.0`. Positive values penalize new tokens based on whether they already appear in the text, thus increasing the likelihood of the model to talk about new topics.	Number	FALSE
Frequency Penalty	A number from `-2.0` through `2.0`. Positive values penalize new tokens based on whether they already appear in the text, thus decreasing the likelihood of the model to repeat the same line.	Number	FALSE
User	A unique identifier representing your end user. OpenAI can use this ID to monitor and detect abuse. See Safety Best Practices.	Text	FALSE

Output

Field	Definition	Type
Model	The model used in the chat completion.	Text
ID	A unique identifier for the chat completion chunk	Text
Object	The object type. This is always `chat.completion`.	Text
Created	The Unix timestamp when OpenAI created the chat completion. The value is in seconds.	Number
Choices	A list of the chat completion choices: `index`: The index of the choice in the list of choices. `message`: A chat completion message generated by the model. `role`: The role of the author of the message. `content`: The contents of the message. `finish reason`: The reason that the model stopped generating tokens. See the chat completion API for a complete description of these fields. This result can be more than one, if N is greater than `1`.	List of Objects
Usage	The usage statistics for the completion request: `prompt tokens`: Number of tokens in the prompt. `completion tokens`: Number of tokens in the generated completion. `total tokens`: Total number of tokens used in the request. This is the sum of the previous two values.	Object
Raw Output	The raw output returned by the chat.	Object

Example

The following is an example of the Raw Output result returned by the card:

{
  "id": "chatcmpl-123",
  "object": "chat.completion",
  "created": 1677652288,
  "model": "gpt-3.5-turbo-0613",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "\n\nHello there, how may I assist you today?"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 9,
    "completion_tokens": 12,
    "total_tokens": 21
  }
}