Update to the latest OpenAI OpenAPI Specification

[AbQaadir]

Overview

This pull request introduces significant updates to the ChatCompletionRequest specifications, enhancing the API's capabilities and providing more flexible options for developers.

Key Changes

New Models

• Added support for new model variants:
  • o1-preview and o1-preview-2024-09-12
  • gpt-4o-realtime-preview
  • gpt-4o-audio-preview
  • chatgpt-4o-latest

New Parameters

1. Store Parameter

   • Added store (boolean) to optionally store model responses
   • Default is set to false
   • Enables persistent response storage for later retrieval

2. Metadata Support

   • Introduced metadata parameter
   • Allows adding custom metadata to requests
   • Improves request tracking and categorization

3. Enhanced Token Management

   • Deprecated max_tokens
   • Introduced max_completion_tokens for more precise output token control

4. Modalities

   • Added modalities parameter
   • Supports specifying interaction types (text, image, audio)

5. Prediction and Audio

   • New prediction parameter for advanced response predictions
   • Added audio parameter to handle audio input/output
   • Enables more sophisticated multimodal interactions

Response Format Improvements

• Expanded response_format to support:
  • JsonObject
  • JsonSchema
  • Provides more structured data handling options

Service Tier

• Explicitly set default service_tier to "auto"

Benefits

• Enhanced multimodal capabilities
• Improved response management
• Better support for diverse use cases

Specifications Comparison ChatCompletionRequest

Parameters	Old Specifications	Updated Specifications	Differences and Additional Information
model	gpt-4o, gpt-4o-2024-05-13, gpt-4o-mini, gpt-4o-mini-2024-07-18, gpt-4-turbo, gpt-4-turbo-2024-04-09, gpt-4-0125-preview, gpt-4-turbo-preview, gpt-4-1106-preview, gpt-4-vision-preview, gpt-4, gpt-4-0314, gpt-4-0613, gpt-4-32k, gpt-4-32k-0314, gpt-4-32k-0613, gpt-3.5-turbo, gpt-3.5-turbo-16k, gpt-3.5-turbo-0301, gpt-3.5-turbo-0613, gpt-3.5-turbo-1106, gpt-3.5-turbo-0125, gpt-3.5-turbo-16k-0613	o1-preview, o1-preview-2024-09-12, o1-mini, o1-mini-2024-09-12, gpt-4o, gpt-4o-2024-11-20, gpt-4o-2024-08-06, gpt-4o-2024-05-13, gpt-4o-realtime-preview, gpt-4o-realtime-preview-2024-10-01, gpt-4o-audio-preview, gpt-4o-audio-preview-2024-10-01, chatgpt-4o-latest, gpt-4o-mini, gpt-4o-mini-2024-07-18, gpt-4-turbo, gpt-4-turbo-2024-04-09, gpt-4-0125-preview, gpt-4-turbo-preview, gpt-4-1106-preview, gpt-4-vision-preview, gpt-4, gpt-4-0314, gpt-4-0613, gpt-4-32k, gpt-4-32k-0314, gpt-4-32k-0613, gpt-3.5-turbo, gpt-3.5-turbo-16k, gpt-3.5-turbo-0301, gpt-3.5-turbo-0613, gpt-3.5-turbo-1106, gpt-3.5-turbo-0125, gpt-3.5-turbo-16k-0613	New models introduced "o1-preview", "o1-preview-2024-09-12", "gpt-4o-realtime-preview", "gpt-4o-audio-preview", "chatgpt-4o-latest".
store	Not present	boolean? store = false;	New parameter. Allows storing model responses for later use. Default value is false.
metadata	Not present	record { string...; }? metadata?;	New parameter. Developers can add custom metadata to requests, allowing better tracking or categorization.
max_tokens	int? max_tokens?;	@deprecated int? max_tokens?;	Deprecated. max_tokens is replaced with max_completion_tokens in the new version to provide more flexibility.
max_completion_tokens	Not present	int? max_completion_tokens?;	New parameter. Specifies a hard limit on the number of tokens for model output, replacing max_tokens.
modalities	Not present	ChatCompletionModalities? modalities?;	New parameter. Specifies the interaction modalities (e.g., text, image, or audio).
prediction	Not present	PredictionContent? prediction?;	New parameter. Specifies the content prediction for responses, enabling more advanced predictions.
audio	Not present	CreateChatCompletionRequest_audio? audio?;	New parameter. Handles audio input/output, enabling multimodal features.
response_format	CreateChatCompletionRequest_response_format response_format?;	ResponseFormatText ResponseFormatJsonObject ResponseFormatJsonSchema response_format?;	Updated. Supports new response formats (JsonObject, JsonSchema) for structured data handling.
service_tier	"auto" "default"? service_tier?;	"auto" "default"? service_tier = "auto";	Default value. Explicitly sets the default value to "auto"

Resources and References

Official Documentation

• OpenAI Chat Completion API Reference
• OpenAPI Specification

[CLAassistant]

All committers have signed the CLA.

[xlight05]

This is great, As we discussed offline, even though from openai did not change the version in this case, it seems like a breaking change according to semver tool. So shall we bump the version to 4.0.0?

[AbQaadir]

Thank you! I’ve added some minor documentation to the code and bumped the version to 4.0.0 as discussed.

[xlight05]

LGTM