On Thursday, September 12th, Artificial Intelligence research and deployment company OpenAI announced the release of their new series of AI models, o1. Designed to reason through complex tasks and solve harder problems than previous AI models such as the company’s flagship GPT-4, o1 is trained to spend more time thinking through problems before coming up with a response.
How Does OpenAI o1 Work?
According to OpenAI, o1 performs similarly to PhD students in scientific benchmark tests and excels in both coding and math compared to previous models. GPT-4o, for instance, was tested against a qualifying exam for the International Mathematics Olympiad and scored just 13%, but the new model raises the bar significantly with an 83% score.
This significant increase in performance is what prompted OpenAI to reset their version counter back to one, and class what was previously known as Project Strawberry as a new series entirely rather than a continuation of their existing AI model.
OpenAI has launched the o1 series with two models, the o1-preview and the o1-mini. As the names suggest, o1-mini is a more streamlined and therefore less powerful version, optimized for coding and STEM tasks and offering faster but also cheaper reasoning ability. 80% cheaper, in fact, with only a slightly reduced performance score across math and coding benchmarks.
Is OpenAI o1 Better Than GPT?
All of OpenAI’s technical research indicates that the o1 series is a drastic improvement on GPT, including from a safety standpoint. GPT is comparatively vulnerable to having its safety protocols bypassed by what is called “jailbreaking”, but o1 has demonstrated a significant improvement on this score.
OpenAI have also bolstered their own safety standards and evaluations, and formalized agreements with the AI Safety Institutes of both the United States of America and the United Kingdom. Both institutes have been granted early access to a research version of o1.
Should You Use OpenAI o1?
Yes, but currently only if you’re needing complex scientific problems solved. Geneticists, physicists, and developers among others can all enjoy significant mileage from o1 right away, but it’s not going to be the right model for uploading files or trawling the web for information like ChatGPT does.
According to early adopters of the new model, o1 is already the number one choice for certain applications—the deeper thinking and large output token limits of o1-mini, for instance, allow it to perform big jobs in one go compared to existing models which are more suited to a larger number of small tasks.
If you do have tasks that can benefit from the problem-solving power of this new breakthrough in AI, you’ll be able to query it right away if you’re a ChatGPT Plus or Team user. Currently, you’re limited to 50 messages per week for o1-preview and 50 per day for o1-mini.
ChatGPT Enterprise and Edu users will also have access to both models, and developers who qualify for API usage tier 5 can also use both models with a rate limit of 20 RPM at the time of writing. These limits will likely be revised upward in the future, and OpenAI has announced that ChatGPT Free users can look forward to o1-mini access in the future.
How Can You Use AI?
AI has been a source of endless hype since OpenAI’s introduction of GPT-4 in early 2023. Generative AI has had its share of justified critics since then, and while it feels like there’s been a scramble from every boardroom and marketing team to get on the bandwagon, there’s been plenty of money made by people who’ve been able to crack the code.
Entrepreneurs have been leveraging generative AI to create alternate revenue streams with different AI side hustles, and the technology has also helped entrepreneurs to sell more products.
Find AI Advice, Education, and Tools on Whop
If you want to learn how to take advantage of AI by building a business of your own, or get ahead of this new AI model in the o1 series, head over to Whop. Under the AI Life Hacks and Business categories, you’ll find entrepreneurs and communities who are already using AI to enhance their businesses and earn income, and you can bet that o1 is already being dissected for its creative potential.
