OpenAI, the organization that brought ChatGPT to the public two years ago, is hosting a period referred to as “shipmas.” Kicked off on December 6, the AI company will reveal new products, features, and demonstrations for twelve days straight. The company is already through eight days of the Shipmas event and in this story, we are summarizing the highlights of these eight days.
Day 1
OpenAI has officially launched the OpenAI o1 model for ChatGPT, alongside a new subscription option for businesses priced at $200, named ChatGPT Pro. The OpenAI o1 model enhances the user experience by breaking down inquiries and tasks into manageable components, addressing each one before delivering a comprehensive response. The company asserts that this model significantly improves the accuracy and relevance of ChatGPT’s answers. It has demonstrated the capability to solve 83% of questions from the International Mathematics Olympiad while exhibiting fewer errors compared to its predecessor. Additionally, response times have improved. The OpenAI o1 model also introduces an image processing feature, allowing users to upload photos for analysis, explanation, or even the creation of guides based on the content. Click here to experience OpenAI o1.
ALSO SEE: OpenAI Releases ChatGPT Pro, A $200 Monthly Subscription Model With Maxed Out Capabilities
Day 2
The second day was dedicated to developers, highlighting the launch of AI Reinforcement Fine-Tuning. This technique enhances the performance of AI models by refining their reasoning capabilities for generating responses. RFT serves as a valuable resource for developers aiming to tailor OpenAI’s algorithms to meet the demands of specific tasks, particularly those that are intricate. Its application could significantly benefit AI tools utilized in fields such as law and science.
Day 3
OpenAI has made a significant announcement today with the launch of Sora, its AI image generation tool. Initially teased a year ago, Sora is now available on its website for ChatGPT Plus and Pro members. This AI video creator utilizes an enhanced model known as Sora Turbo, which delivers faster and higher-quality video production compared to previous versions. Users can generate videos from text prompts, animate still images, or remix existing videos based on text input. The service operates on a credit system, providing subscribers with 1,000 monthly credits, allowing for the creation of 50 videos at 720p, each lasting five seconds. Subscribers who opt for the $200 monthly plan can produce videos up to 20 seconds long at 1080p resolution. For more information, visit Sora.com.
ALSO SEE: Is OpenAI Planning To Launch A Web Browser? Here’s What We Know
Day 4
OpenAI has introduced an enhancement to ChatGPT known as Canvas, designed to facilitate collaboration with both code and text. This feature allows users to view responses from the AI chatbot on a shared, editable interface. With Canvas, individuals can collaborate with ChatGPT on various writing and coding tasks. Developers can utilize Canvas for code editing, as the AI can debug, suggest improvements, and provide comments on the code similarly to how it interacts with text. Additionally, Canvas has the capability to edit and translate code into Python. Currently, Canvas is available in its first non-beta version for ChatGPT users.
Day 5
OpenAI revealed the integration of ChatGPT into Apple Intelligence on the fifth day, launching it as part of iOS 18.2. This integration aims to enhance Siri’s capabilities and supports various new features introduced by Apple on that same day. With ChatGPT integrated into Apple Intelligence, users can expect more nuanced answers to complex inquiries, moving away from mechanical responses to more contextual interactions. However, the announcement was overshadowed by an outage at OpenAI, which rendered ChatGPT, Sora, and APIs inoperable. Fortunately, the service has since been restored and is functioning normally.
Day 6
At the midpoint of its development, OpenAI introduced Advanced Voice Mode for ChatGPT, which includes visual capabilities. This feature allows users to share their screens and stream video directly to the AI chatbot. Instead of uploading images or screenshots to pose questions, ChatGPT can now observe your view and provide guidance. It resembles a video chat with a friend, although in this scenario, the friend does not reciprocate with their own visuals. Users can listen to the AI’s voice, and the interaction is entirely hands-free.
ALSO SEE: Here’s How OpenAI’s ChatGPT Can Help Catch Students Who Are Cheating in Exams
Day 7
OpenAI dedicated the seventh day to enhancing organization. ChatGPT introduced Projects, a new feature designed to streamline your ChatGPT sessions. Projects allow you to categorize your interactions with ChatGPT, including any associated files and data. This feature acts like a virtual filing cabinet, complete with neatly labeled folders curated by an assistant familiar with the contents. When initiating a project, you must assign it a name and a color for easier identification. The following step involves grouping related conversations and uploading pertinent files. You can either incorporate existing chats or initiate new ones. With the addition of functionalities like SearchGPT and Canvas, Projects will operate even more efficiently.
ALSO SEE: Apple Could Charge For Exclusive AI Features, To Get An Observing Seat At OpenAI Later This Year
Day 8
The organization has introduced its conversational search capability, ChatGPT Search, to all ChatGPT accounts and has integrated it into Advanced Voice Mode. This search function will operate with enhanced precision, allowing the AI to determine whether to retrieve up-to-date information from the internet or to respond using its existing knowledge. The results will feature web previews, images, and, when appropriate, videos.
Day 9
The ninth day of the Shipmas event saw OpenAI announcing OpenAI o1 in the API with support for developer messages, function calling, vision capabilities, and Structured Outputs. The key features of OpenAI o1 are ready now and they enable real-world use cases that include seamlessly connecting o1 to external data and APIs, generating responses that adhere to custom JSON schema, specifying instructions or context for the model to follow and reasoning over images to unlock many more apps in science. It uses 60% fewer reasoning tokens and is faster than before.
Day 10
The tenth day of the Shipmas event saw OpenAI announcing a direct call and WhatsApp messaging support for ChatGPT in the US. This new feature allows users to connect with the chatbot by calling the number 1-800-ChatGPT or sending a WhatsApp message to the same number, available in any region where ChatGPT is accessible. It is an experimental feature that has been made to offer access to ChatGPT without making an account. Those above 13 years of age can use the service and calls can be placed from flip phones and landlines too. Users can talk to ChatGPT for up to 15 minutes per month for free and messages can be sent daily, but there is a fixed limit. The service is currently available in the US for phone calls and globally for WhatsApp messaging wherever ChatGPT is available.
Day 11
The eleventh day of the Shipmas event marked OpenAI’s announcement of enhanced support for ChatGPT, enabling it to integrate with various coding and note-taking applications on macOS via text or voice commands. This Beta feature is available to Plus, Pro, Team, Enterprise, and Edu users, aiming to optimize workflow by embedding ChatGPT more thoroughly into the desktop environment. The Work with Apps functionality allows the AI to engage directly with compatible applications, including coding text editors. Supported apps encompass Apple Notes, Notion, Quip, Xcode, and VS Code. This feature has been launched on macOS today, with plans to extend it to Windows and Free users in 2025.
Day 12
The last day of the Shipmas event saw OpenAI announcing its latest reasoning models namely the o3 and o3 Mini. These models were showcased as a preview and there are no plans for public release yet. However, OpenAI CEO Sam Altman has claimed that the o3 Mini model will launch by the end of January 2025 and the full version of the o3 model will follow soon after. These advanced LLM models have the capacity to outperform their predecessors and thus attract new users and many more investors.
You can follow Smartprix on Twitter, Facebook, Instagram, and Google News. Visit smartprix.com for the latest tech and auto news, reviews, and guides.
