Create Chat Completion

User Interface

Dataflow

Field	Description	Comment
Model	Select GPT Model
Messages	A list of messages comprising the conversation so far.	Refer to "How to make messages?" section below this page.
Store	Whether or not to store the output of this chat completion request for use in our model distillation or evals products.	Default: false
Reasoning Effort	Constrains effort on reasoning for reasoning models. (Optional)	low, medium(default), high o-series models only
Metadata	Set of 16 key-value pairs that can be attached to an object. (Optional)	Keys are strings with a maximum length of 64 characters. Values are strings with a maximum length of 512 characters. Refer to "What is Metadata?" section below this page.
Frequency penalty	An optional parameter that controls repetition of the same words or phrases. (Optional)	Allow Range : -2.0 ~ 2.0 (Default : 0)
Logic bias	Modify the likelihood of specified tokens appearing in the completion. (Optional)	Refer to "What is Logic Bias?" section below this page.
Max Completion Tokens	An upper bound for the number of tokens that can be generated for a completion, including visible output tokens and reasoning tokens. (Optional)
N	The number of responses to generate per request. (Optional)	A positive integer (default is 1)
Precense penalty	A parameter that encourages the model to explore topics not previously mentioned. (Optional)	Allow Range : -2.0 ~ 2.0 (Default : 0)
Response Type	Select Response Type (Optional)	Text: Returns the response as a plain string (Default) JSON Object: Returns the response as a JSON object JSON Schema: Returns the response in the specified JSON Schema format
JSON Schema	Setting to receive the response as a desired JSON object. (Optional)	Required when specifying the response data type as JSON Schema. Refer to "What is JSON Schema?" section below this page.
Service Tier	Specifies the latency tier to use for processing the request. This parameter is relevant for customers subscribed to the scale tier service (Optional)	Auto(Default) Default
Stop	Up to 4 sequences where the API will stop generating further tokens. The returned text will not contain the stop sequence. (Optional)	Use commas (,) to separate multiple settings Example: stop,halt
Temperature	A parameter that controls randomness during the model's response generation (Optional).	Allow Range : 0.0 ~ 2.0 (Default 1)
Top P	Limits sampling to only the top-p portion of the probability distribution during response generation. (Optional)	Allow Range : 0.0 ~ 1.0 (Default 1)
Tools	Specifies the list of functions the model is allowed to call (Optional).	Refer to "What is Tools?" section below this page.
Tool Choice	Optional setting to determine whether the model should call a tool. (Optional)	Refer to "What is Tool Choice?" section below this page.
Tool Choice Option	Setting to enable or disable parallel tool calls. (Optional)	None (default) Yes No
User	Setting to assign a unique ID for identifying the end user. (Optional)

How to make messages?

The message content is structured as an array of message objects, and each message follows the format below.

Key	Description	Comment
content	Message Content
role	Message Role	system: Defines the purpose or behavior of the AI model for the conversation. user: Represents the message or question sent by the user to the AI—this is the actual input the AI should respond to. assistant: Represents the AI model’s response to the user’s input. developer: Provides additional information to the AI model from the developer.

Message Context Example:

[
  {
    "role": "system",                          // Defines the nature of the conversation
    "content": "You assist users in creating workflows."
  },
  {
    "role": "assistant",                       // AI's sample response
    "content": "Hello! How can I help you today?"
  },
  {
    "role": "user",                            // User request to the AI
    "content": "Could you explain what an automation workflow is, in English?"
  },
  {
    "role": "developer",                       // Developer's instruction to guide the conversation
    "content": "The user wants an explanation of a workflow in English. Please translate accordingly."
  }
]

What is Metadata?

A field provided to allow users to store additional information related to the API request.

Example:

{
  "user_id": "c2453689-9888-4222-b870-fb36cbd8fa93" // Identified by an opaque identifier.
}

What is Logic Bias?

A feature that allows you to forcibly adjust the probability of specific words (or tokens) appearing.

Example:

// To increase the likelihood of the word "Cat" appearing more often
// The token ID for "Cat" is 9240
{
    "9240": 50
}
// Value range: -100 to 100
// Positive values increase repetition of the word
// Negative values suppress or block the word

What is JSON Schema?

A setting that emphasizes the model to respond in a specific JSON format.

Example:

{
  "type": "json_schema",
  "json_schema": {
    "type": "object",
    "properties": {  // Defines the structure and data types of the response
      "temperature": { "type": "number" },
      "condition": { "type": "string" },
      "humidity": { "type": "integer" },
      "city": { "type": "string" }
    },
    "required": ["temperature", "condition", "city"] // Specifies required fields in the response
  }
}


// Example response based on the defined JSON Schema: 
{
  "temperature": 12.5,
  "condition": "Sunny",
  "humidity": 60,
  "city": "Seoul"
}

What is Tools?

A feature that allows the model to call specific functions. In other words, the model can generate JSON directly to invoke APIs or interact with external systems.

Example:

// Example tool setup to allow the model to fetch current weather for a specific city

[
  {
    "type": "function",
    "function": {
      "name": "get_weather",
      "description": "Fetches the current weather for a specific city.",
      "parameters": {
        "type": "object",
        "properties": {
          "city": { "type": "string", "description": "Name of the city to retrieve the weather for" },
          "unit": { "type": "string", "enum": ["metric", "imperial"], "description": "Temperature unit (metric = Celsius, imperial = Fahrenheit)" }
        },
        "required": ["city"]
      }
    }
  }
]


// Example output in a conversation using the tool

{
  "model": "gpt-4-turbo",
  "messages": [
    { "role": "system", "content": "You are an AI that provides weather information." },
    { "role": "user", "content": "Tell me the weather in Seoul." },
    {
      "role": "tool",
      "tool_call_id": "123456",
      "name": "get_weather",
      "arguments": {
        "city": "Seoul",
        "unit": "metric"
      }
    },
    {
      "role": "assistant",
      "content": "The current weather in Seoul is clear, with a temperature of 12°C."
    }
  ]
}

What is Tool Choice?

An option that determines how the model chooses to use a tool.

Options:

Option	Description
auto	The model will call the tool if needed, and will not call it if unnecessary. (Default setting)
none	The model will never call a tool. Only text responses are allowed.
required	The model must call at least one tool.
{ "type": "function", "function": { "name": "my_function" } }	Forces the model to call a specific tool (my_function).

---

Response Datas

Key	Description
id	Unique identifier of the response request
object	The type of the response object, indicating a conversational reply from GPT.
created	The Unix timestamp (in seconds) when the response was generated.
model	The name of the GPT model used to generate the response.
system_fingerprint	A unique system fingerprint used for internal logging and tracking.
choices[].index	The index of the response, used to distinguish between multiple responses in a single request.
choices[].message.role	The role of the entity that generated the message.
choices[].message.content	The content of the generated response.
choices[].message.refusal	Indicates whether the request was denied.
choices[].logprobs	Log probabilities used during response generation.
choices[].finish_reason	The reason why the response generation was stopped.
usage.prompt_tokens	The number of tokens used in the input prompt.
usage.completion_tokens	The number of tokens used in the generated response (completion).
usage.total_tokens	The total number of tokens used, including both the input prompt and the generated response.
usage.prompt_tokens_details.cached_tokens	The number of cached prompt tokens.
usage.completion_tokens_details.reasoning_tokens	The number of tokens used for reasoning during response generation.
usage.completion_tokens_details.accepted_prediction_tokens	The number of predicted tokens accepted by the model during response generation.
usage.completion_tokens_details.rejected_prediction_tokens	The number of predicted tokens rejected by the model during response generation.