For the past few weeks, OpenAI has been relatively quiet in terms of new product announcements and updates. They are coming out of this brief lull period with a series of new announcements starting tomorrow.
12 days.
— OpenAI (@OpenAI) December 4, 2024
12 livestreams.
A bunch of new things, big and small.
12 Days of OpenAI starts tomorrow.
The event, dubbed '12 Days of OpenAI,' is starting tomorrow at 10 am Pacific Time. Each weekday, OpenAI will host a live stream with a launch or demo. OpenAI also mentioned that this series of announcements will include some big ones and some small ones. OpenAI CEO Sam Altman tweeted that they have some great stuff to share.
🎄🎅starting tomorrow at 10 am pacific, we are doing 12 days of openai.
— Sam Altman (@sama) December 4, 2024
each weekday, we will have a livestream with a launch or demo, some big ones and some stocking stuffers.
we’ve got some great stuff to share, hope you enjoy! merry christmas.
Here's what we are expecting from OpenAI as part of the upcoming series of announcements:
Full o1 model: Back in September, OpenAI announced the preview of the o1 series of models that are better suited for complex tasks. The o1 preview models were able to solve harder problems than previous models in science, coding, and math. While the o1 preview model remains impressive, OpenAI may release the full o1 model. You can find more details on what to expect from the full o1 model here.
Sora public preview: Early this year, OpenAI announced Sora, its video generation model. Due to various reasons, OpenAI delayed the public preview of Sora. OpenAI's former chief technology officer, Mira Murati, earlier revealed that their plan is to launch Sora to the general public sometime later in 2024. So, the Sora preview may come this week.
ChatGPT Video Input: When OpenAI launched the GPT-4o model, they demoed both the Advanced Voice Mode and the ability to process real-time video input. While the Advanced Voice Mode is already available for all ChatGPT paid tier users, the real-time video input capability never saw the light of day. So, OpenAI may finally announce real-time video input support in ChatGPT in the coming days.
Increased context length: When compared to Google Gemini and other recent models, OpenAI's models are severely lacking in the context window arena. OpenAI may finally increase their context window of 128,000 tokens for their latest models.
New GPT model (GPT-4.5 or GPT-5): Back in May, when the GPT-4o model was launched, it was a clear leader in the LLM category. It delivered state-of-the-art results in most of the key AI benchmarks. Since then, LLMs from Google, Alibaba, Anthropic, Amazon, and others have caught up with the performance capabilities of GPT-4o. In fact, GPT-4o is now lagging behind its rivals in several key categories, including coding and math. So, OpenAI may announce its next big model update in the coming weeks.
Dall-E updates: Like GPT-4o, competitors have surpassed the capabilities of the Dall-E 3 image generation model, which was launched back in 2023. OpenAI may finally announce the next major update to its image generation model to coincide with the launch of the Sora video generation model.
Agents: OpenAI CEO Sam Altman has been mentioning agents for a while now. OpenAI may announce new capabilities that will allow developers to easily build agents using OpenAI's models.
Apart from the above, OpenAI may also announce improvements to ChatGPT apps, wider availability of ChatGPT web search capabilities, and new tools and improvements for developers to build better AI applications.
The expectations from consumers and developers are high for OpenAI; hopefully, they will deliver.
Image Credit: Kevin Weil
1 Comment - Add comment