In summary
- OpenAI is about to launch two revolutionary models, Strawberry and Orion, that could redefine the machine learning landscape.
- Strawberry, formerly known as Q* or Q-Star, focuses on significant advances in reasoning and solving complex mathematical problems.
- Orion, OpenAI’s next flagship language model, is designed to outperform GPT-4 in language understanding and generation, as well as processing multimodal inputs such as text, images, and videos.
OpenAI is on the cusp of releasing two revolutionary models that could redefine the machine learning landscape. Codenamed Strawberry and Orion, these projects aim to push AI capabilities beyond current limits, especially in reasoning, problem solving, and language processing, bringing us one step closer to artificial general intelligence (AGI).
Strawberry, formerly known as Q* or Q-Star, appears to be more than just a chatbot; it is focused on showing significant advancement in the reasoning abilities of artificial intelligence. Sources familiar with the project have reported to different media such as Reuters or The Information that it has demonstrated remarkable competence in solving complex mathematical problems and in logical analysis.
Meanwhile, Orion is shaping up to be OpenAI’s next flagship language model, with the potential to replace GPT-4. It is designed to outperform its predecessor in language understanding and generation, while also incorporating the ability to process multimodal inputs, such as text, images, and videos.
Both projects have attracted the attention of US national security officials, underscoring their potential strategic importance. This development comes as OpenAI continues to raise capital despite considerable revenue growth, likely due to the high costs associated with developing and training these advanced models.
Strawberry and reasoning power
Despite an endless flood of online speculation, OpenAI has not officially said anything about Project Strawberry. However, the alleged leaks tend towards its capabilities for sophisticated reasoning.
Unlike traditional models that provide quick answers, Strawberry is said to employ what researchers call “System 2 thinking,” capable of taking the time to deliberate and reason about problems, rather than predicting longer sets of tokens to complete its answers. This approach has yielded impressive results, with the model scoring over 90% on the MATH benchmark (a collection of advanced math problems), according to Reuters.
Another key innovation anticipated from Strawberry is its ability to generate high-quality synthetic training data. This addresses a critical challenge in AI development: the scarcity of diverse, high-quality data to train models on. If true, Strawberry not only improves its own capabilities, but also paves the way for more advanced models like Orion.
Considering the massive amounts of data already collected by OpenAI, and the privacy movement now rife among users unwilling to give their data to AI trainers, this feature may play a major role in the quality of future AI models, much like some users today train their own custom models using images generated by Stable Diffusion.
However, Strawberry’s deliberate approach to processing can present challenges for real-time applications. OpenAI researchers are reportedly working on “distilling” Strawberry’s capabilities, essentially lowering its quality so that consumers can perform large amounts of inferences at low computational cost.
Still, the potential integration of Strawberry’s technology into consumer-facing products like ChatGPT could mark a significant boost in how OpenAI trains new models. However, it’s possible that OpenAI will use Strawberry as a foundation for training new models rather than making it widely available to consumers.
Project Orion or GPT Next
Project Orion is billed as the ambitious successor to OpenAI’s GPT-4o, aiming to set new standards in language AI. A recent presentation by Tadao Nagasaki, CEO of OpenAI Japan, suggests it could be called GPT Next. Building on the advancements of Project Strawberry, Orion is designed to excel at natural language processing while expanding into multimodal capabilities.
And OpenAI says the leap won’t be incremental.
“The next AI model, likely called ‘GPT Next,’ will evolve nearly 100 times more than its predecessors, based on past performance,” Nagasaki said at the KDDI SUMMIT 2024 in Japan, IT Media reported. “Unlike traditional software, AI technology grows exponentially. Therefore, we want to support the creation of a world where AI is integrated as soon as possible.”
Training Orion on data produced by Strawberry would represent a technical advantage for OpenAI. However, this technique should be used with caution. Researchers have already shown that models begin to degrade after being trained on too much synthetic data, so finding that sweet spot where Strawberry can boost Orion without affecting its accuracy seems key for OpenAI to remain competitive.
Orion’s native multimodal capabilities will also represent a significant advancement. The model is being developed to seamlessly integrate text, image, and even video input and output, The Information reported, opening up new possibilities for ChatGPT users and putting the company in direct competition with Google’s Gemini, which can process up to 2 hours of video input.
This is the model that users will interact with when using ChatGPT or the OpenAI Playground API.
The development of Orion aligns with OpenAI’s broader strategy of maintaining its competitive edge in an increasingly crowded AI landscape. With open-source models like Meta’s LLaMA-3.1, and cutting-edge models like Claude or Gemini advancing rapidly, Orion is basically OpenAI’s attempt to stay ahead of the curve.
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
For the Latest Crypto News, Follow ©KeynoteUSA on Twitter Or Google News.
Disclaimer: Please note that the information provided on this page is for News purposes only and should not be considered investment or trading advice. ©Crypto.keynoteusa.com strongly recommends that you conduct independent research and/or consult with a qualified professional before making any investment decisions.
