Open AI’s 12 Days of Shipmas
OpenAI's "12 Days of OpenAI" event, colloquially known as "shipmas," commenced on December 5, 2024, unveiling a series of innovations to celebrate the second anniversary of ChatGPT, which now serves over 300 million weekly users. Below is a comprehensive recap of the announcements made during this event:
Day 1: December 5, 2024
Release of OpenAI o1 Reasoning Model: OpenAI introduced the full version of its o1 reasoning model, enhancing capabilities in complex problem-solving, including coding, mathematics, and scientific research. ChatGPT Plus and Team users received immediate access, with Enterprise and Edu users scheduled to gain access the following week.
Introduction of ChatGPT Pro Subscription: A new premium subscription tier, ChatGPT Pro, was introduced at $200 per month. This tier offers unlimited access to GPT-4o, Advanced Voice Mode, and an exclusive, more advanced version of the o1 model, referred to as o1 pro mode. The Pro subscription is designed for power users requiring enhanced computational power for complex tasks.
Day 2: December 6, 2024
OpenAI's Reinforcement Fine-Tuning Research Program: OpenAI expanded alpha access to its Reinforcement Fine-Tuning program, inviting researchers, universities, and enterprises with complex tasks to apply. This initiative allows users to fine-tune the o1 model on their own datasets, leveraging OpenAI's reinforcement-learning algorithms to customize domain-specific AI models. The technology is expected to be publicly available next year, enabling a broader range of applications across various fields.
Day 3: December 9, 2024
Launch of Sora, the Text-to-Video AI Generator: OpenAI officially launched Sora, a text-to-video AI tool that transforms textual descriptions into realistic and imaginative videos. Sora has been in development and testing phases, with artists participating in an invite-only research preview. Its release is expected to significantly impact the generative AI video domain.
Day 4: December 10, 2024
Introduction of Canvas: OpenAI unveiled Canvas, a new feature that allows users to collaborate with ChatGPT in a side-by-side digital editing environment. Canvas is designed to enhance writing and coding tasks by providing an interactive space for real-time collaboration. This feature is now available to all web users, both free and paid.
Day 5: December 11, 2024
ChatGPT Integration with Apple Intelligence: Timed with the release of iOS 18.2, OpenAI announced the integration of ChatGPT with Apple's ecosystem. This integration enables users to access ChatGPT's capabilities seamlessly across Apple devices, enhancing user experience and accessibility.
Day 6: December 12, 2024
Advanced Voice Mode Enhancements: OpenAI introduced improvements to ChatGPT's Advanced Voice Mode, including the ability to analyze visual inputs through a mobile device's camera or screen sharing. This enhancement allows users to ask questions about their surroundings without needing to describe the scene or upload photos, providing a more intuitive interaction experience.
Day 7: December 13, 2024
Launch of ChatGPT Projects: OpenAI introduced "Projects," a folder system designed to help users better organize their work with ChatGPT. This feature allows users to create new projects, customize titles and icons, and add files and instructions to personalize ChatGPT's responses. Projects are being rolled out to ChatGPT Plus, Pro, and Teams users immediately, with availability for free users targeted "as soon as possible" and for Enterprise and Edu users early in the new year.
Day 8: December 16, 2024
SearchGPT Availability: OpenAI announced that its AI search engine, SearchGPT, is now available to all users, both free and paid. Initially released in October for paid subscribers, the search engine now requires only a logged-in account, even at the free tier. Improvements for mobile users include a more traditional search engine appearance, with location-based searches featuring lists of results, images, ratings, hours, and detailed information, including maps and directions directly within the app.
Day 9: December 17, 2024
Santa Mode Introduction: As part of the holiday season, OpenAI launched "Santa Mode," allowing users to speak directly with an AI representation of Father Christmas. This feature is available on mobile, web, and desktop versions of ChatGPT and can be accessed by selecting a snowflake icon next to the prompt bar. Santa Mode is intended primarily for young families with children and will be available until early January.
Day 10: December 18, 2024
1-800-CHATGPT Service: OpenAI launched a new feature allowing users to call ChatGPT using the phone number 1-800-CHATGPT for up to 15 minutes per month at no cost. This service aims to make AI more accessible to newcomers by providing a simplified version of ChatGPT via a familiar channel. The phone service uses OpenAI's Realtime API, while the WhatsApp functionality employs GPT-4o mini through the WhatsApp API.
Day 11: December 19, 2024
Realtime API Updates: OpenAI announced updates to its Realtime API, including simple WebRTC integration, a 60% price reduction for GPT-4o audio, and support for GPT-4o mini at one-tenth of previous audio rates. These updates aim to make real-time AI
Day 12: December 20, 2024
o3 and o3-mini Reasoning Models
On the final day of “shipmas,” OpenAI announced the upcoming release of two advanced reasoning models, o3 and o3-mini, set to debut in early 2025. These models represent the next step in AI reasoning capabilities, improving performance in coding, mathematics, and scientific research. The o3-mini version introduces flexibility, allowing users to adjust reasoning levels to balance performance and computational cost.
Additionally, OpenAI unveiled a new safety strategy called “deliberative alignment,” leveraging the enhanced reasoning abilities of these models to identify unsafe prompts and establish safety protocols. Public testing for these models will begin shortly, with OpenAI inviting researchers to evaluate their capabilities and provide feedback.
Conclusion
The “12 Days of OpenAI” highlighted the company’s ongoing commitment to pushing the boundaries of generative AI. From enhancing reasoning models and introducing revolutionary tools like Sora and Canvas to expanding accessibility through features like SearchGPT and Santa Mode, OpenAI showcased its ability to innovate across a wide spectrum of applications.
The event not only delivered powerful new tools but also emphasized the importance of user safety and adaptability in AI technology. As OpenAI prepares for 2025 with the launch of o3 and o3-mini, the company’s trajectory underscores its mission to democratize access to AI while maintaining high standards of safety and reliability.
This series has set a high bar for the industry and provided a glimpse into the future of AI-powered innovation.