OpenAI Launches New Developer Features During 'Mini Dev Day' Event

As the holiday season approaches, many companies are launching deals and promotions to engage with customers. OpenAI is also getting into the spirit with its exciting "12 Days of OpenAI" event.

Starting on December 5, OpenAI began a series of daily live streams showcasing a variety of new features and enhancements, both significant and minor, as outlined in a recent announcement from the company.

Understanding the '12 Days of OpenAI'

During a live stream at 10 a.m. PT on December 5, OpenAI CEO Sam Altman elaborated on the twelve-day event, which will feature daily launches or demonstrations. These releases are categorized as either "big updates" or "smaller enhancements," emphasizing the wide range of improvements being shared.

Highlights from the Recent Updates

December 17 - 'Mini Dev Day'

The ninth day of the event put a spotlight on developer-focused features. Key announcements included:

• The introduction of the o1 model, which is now fully available in the API, incorporating features such as function calling and structured inputs along with enhanced vision capabilities and reduced latency.
• A new 'reasoning effort' parameter, enabling developers to adjust how much cognitive effort the model should use in generating responses for better cost efficiency.
• WebRTC support has been added to the Realtime API, facilitating the development of real-time voice applications across different platforms.
• The Realtime API has seen a 60% decrease in audio token prices, added support for GPT-4o mini, and offered enhanced response control.
• Introduction of Preference Fine-Tuning in the fine-tuning API, allowing users to customize model behavior based on preferred and unpreferred responses.
• Beta release of new SDKs for Go and Java.
• An 'Ask Me Anything' session hosted on OpenAI's GitHub platform, available for one hour after the live stream.

December 16 - Improved Search Features

The features revealed on the second Monday centered around enhancements in search capabilities within ChatGPT:

• The AI search engine is now accessible to all users, including those with free accounts.
• The search experience has been made faster with improved mobile functionality, featuring an upgraded map interface with rich visuals.
• Users can now conduct searches verbally through Advanced Voice mode, allowing for a more seamless interaction.
• OpenAI teased developers with a promise that the next day would focus on them, hinting at the upcoming "Mini Dev Day."

December 13 - Introducing 'Projects'

OpenAI responded to user demand for better chat organization by launching the "Projects" feature:

• Projects provide a way to organize and customize chats in ChatGPT.
• Users can create a project with various elements like a title, folder color, relevant files, and specific instructions for ChatGPT.
• Conversations can be saved within the project, allowing for easier access to past discussions.
• This feature is being rolled out to Plus, Pro, and Teams accounts, with plans for free accounts in the future and Enterprise/Education accounts early next year.

December 12 - Advanced Voice Enhancements

In response to prior issues with the live stream, OpenAI quickly shifted focus to a new exciting announcement:

• Advanced Voice Mode was upgraded to include screen-sharing functions.
• These new capabilities enhance user experience by allowing voice interaction while viewing content.
• A new festive "Santa" voice has also been introduced, with special functionality for users.

Additional Announcements

Throughout the previous days, highlights included:

• Integration of ChatGPT with Apple's Siri in the latest iOS update, enhancing the virtual assistant's range.
• Canvas has been made available to all web users, allowing for seamless modifications and integration into various projects.
• The debut of the Sora Turbo video model which significantly expands video capabilities for users.

Accessing Live Streams

Each live stream can be found directly on the OpenAI website, and recordings are uploaded to its YouTube channel shortly after. Links for accessing the streams are shared on OpenAI's social media accounts prior to each event, which occurs daily at approximately 10 a.m. PT.

OpenAI, developer, features, live stream