Open AI, the company that developed the world-famous ChatGPT, has added another model to its impressive lineup of AI models. Over the past few months, the AI developer has released several new models, and their release run is still ongoing.
Related: Open AI Drops New Chat GPT Model With Super-Efficient and Low-Latency ‘GPT-4o mini’
In an official blog post, Open AI introduced a brand new family of AI models to its offerings. According to the company, these new models come with improved performance and reasoning capabilities. These new offerings are now available on Open AI’s ChatGPT Plus subscription tier.
The official name of the new launch is OpenAI o1. It is a new large language model (LLM) specially trained with reinforcement learning to perform complex reasoning. According to the blog post, the new “o1” model is trained to think before it answers any queries.
Many AI industry watchers have been looking forward to Open AI's new “strawberry” model. However, the differences between the numerous OpenAI models under development have not yet been publicly disclosed.
Open AI has, however, described the new family of models as not just another development but a big leap forward worthy of the company changing its usual naming scheme to something newer. So, unlike the previous models named ChatGPT-3, ChatGPT-3.5, and ChatGPT-4.0, this newer model is called “OpenAI o1”.
The reason for the change of name is very significant. OpenAI said the new model is a big advancement and a new level in AI capability, especially for complex reasoning tasks. The improvement is so monumental that they had to “reset the counter back to one and name the series OpenAI o1.”
According to Open AI, one of the most important features of the o1 models is their ability to take their time to think before taking action. This ability gives them an edge over other GPT models, and the company has even affirmed that the smallest o1 model outperforms the powerhouse GPT4.0 in many important areas.
OpenAI revealed that the newly released models emphasize what they call “deliberative reasoning”. The process helps the model produce more thoughtful and coherent answers for reasoning-intensive tasks.
Model o1 has already excelled in the internal testing results released by OpenAI in tasks such as coding, calculus, and even data analysis. But despite all of these, the new model has shown less improvement when it comes to creative tasks such as creative writing. Still, the results generated by o1 were rated well by human evaluators.
You Might Also Like: OpenAI’s Newest Buddy is Atomic Bomb’s Creator, Los Alamos National Lab
Another drawback of the new model is that giving the AI the capacity to think before answering opens the potential risks that stem from heightened intelligence. Of course, that is not to say this model will become sentient; it just has improved rational intelligence.
OpenAI has also said it intends to expand the capabilities of the o1 model. With time, the model will be tweaked to meet general OpenAI minimum standards regarding safety, jailbreak prevention, and autonomy.
Disclaimer: All materials on this site are for informational purposes only. None of the material should be interpreted as investment advice. Please note that despite the nature of much of the material created and hosted on this website, HODL FM is not a financial reference resource and the opinions of authors and other contributors are their own and should not be taken as financial advice. If you require advice of this sort, HODL FM strongly recommends contacting a qualified industry professional.