Skip to content

OpenAIChatCompletionsParameters.md

Latest commit

 

History

History
134 lines (101 loc) · 7.41 KB

OpenAIChatCompletionsParameters.md

File metadata and controls

134 lines (101 loc) · 7.41 KB

OpenAIChatCompletionsParameters

The OpenAIChatCompletionsParameters class is designed to handle the parameters required for chat completions using the OpenAI API.

Inherits

Properties

Property Type Default Value Description
model Text "gpt-4o-mini" ID of the model to use. Supports provider:model aliases for multi-provider usage (e.g., "openai:gpt-4o", "anthropic:claude-3-opus").
stream Boolean False Whether to stream back partial progress. If set, tokens will be sent as data-only. Callback formula required.
stream_options Object Null Property for stream=True. For example: {include_usage: True}
max_completion_tokens Integer 0 The maximum number of tokens that can be generated in the completion.
n Integer 1 How many completions to generate for each prompt.
temperature Real -1 What sampling temperature to use, between 0 and 2. Higher values make the output more random, while lower values make it more focused and deterministic.
top_p Real -1 An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. Only sent when the value is greater than 0 (omitted when <= 0, with default -1).
store Boolean False Whether or not to store the output of this chat completion request.
reasoning_effort Text Null Constrains effort on reasoning for reasoning models. Currently supported values are "low", "medium", and "high".
response_format Object Null An object specifying the format that the model must output. Compatible with structured outputs.
tools Collection Null A list of tools (OpenAITool) the model may call. Only "function" type is supported.
tool_choice Variant Null Controls which (if any) tool is called by the model. Can be "none", "auto", "required", or specify a particular tool.
prediction Object Null Static predicted output content, such as the content of a text file that is being regenerated.
service_tier Text Null Specifies the processing type used for serving the request. "auto", "auto", "default", and "priority".

Asynchronous Callback Properties

| Property | Type | Description | |---------------------------|---------|-------------------------|---------------------------------------------------------------------------------------------------| | onData
(or formula) | 4D.Function | A function to be called asynchronously when receiving data chunk.
Ensure that the current process does not terminate. |

onData will receive as argument a OpenAIChatCompletionsStreamResult

See OpenAIParameters for other callback properties.

Response Format

The response_format parameter allows you to specify the format that the model must output. This is particularly useful when you need structured data or want to ensure the response follows a specific format.

Supported Response Format Types

1. Text Format (Default)

The default response format returns plain text:

var $params := cs.AIKit.OpenAIChatCompletionsParameters.new({ \
    model: "gpt-4o-mini"; \
    response_format: {type: "text"} \
})

2. JSON Object Format

Forces the model to respond with valid JSON:

var $params := cs.AIKit.OpenAIChatCompletionsParameters.new({ \
    model: "gpt-4o-mini"; \
    response_format: {type: "json_object"} \
})

var $messages := [ \
    cs.AIKit.OpenAIMessage.new({ \
        role: "system"; \
        content: "You are a helpful assistant that always responds in JSON format." \
    }) \
]
$messages.push({ \
    role: "user"; \
    content: "Give me information about the number 42 in JSON format with keys 'significance' and 'facts'." \
})

$result := $client.chat.completions.create($messages; $params)

3. JSON Schema Format (Structured Outputs)

For precise control over the JSON structure, you can define a schema:

var $jsonSchema := { \
    type: "object"; \
    properties: { \
        name: {type: "string"}; \
        age: {type: "integer"}; \
        occupation: {type: "string"}; \
        city: {type: "string"} \
    }; \
    required: ["name"; "age"; "occupation"; "city"]; \
    additionalProperties: False \
}

var $params := cs.AIKit.OpenAIChatCompletionsParameters.new({ \
    model: "gpt-4o-mini"; \
    response_format: { \
        type: "json_schema"; \
        json_schema: { \
            name: "person_info"; \
            description: "Information about a person"; \
            schema: $jsonSchema; \
            strict: True \
        } \
    } \
})

var $messages := [{role: "system"; content: "You are a helpful assistant."}]
$messages.push({ role: "user"; content: "Generate information about a person named John Smith."})

$result := $client.chat.completions.create($messages; $params)

JSON Schema Properties

When using json_schema type, you can specify:

  • name: A name for the schema
  • description: A description of what the schema represents
  • schema: The JSON schema definition
  • strict: Whether to enforce strict adherence to the schema

Important Notes

  • Not all models support structured outputs (json_object or json_schema), so check model capabilities before using them.
  • When using json_object format, you should include instructions in your system message to respond in JSON format
  • The json_schema format provides the most control and ensures the response exactly matches your specified structure
  • Invalid JSON responses will result in an error when using JSON formats
  • JSON schema validation ensures type safety and required field presence

See also