OpenAI o1 (ChatGPT o1) Explained: Features and Release
OpenAI o1, formerly Project Strawberry, uses chain-of-thought reasoning and scored 83% on a math olympiad test versus GPT-4o's 13%.
3 min read

In brief
AI summaryOpenAI introduced o1, previously known as Project Strawberry, as a step toward building the world's first AGI, still in preview.
The o1 model scored 83% accuracy on math olympiad qualifying questions, versus just 13% for the earlier GPT-4o model.
o1-preview launched in September 2024 for ChatGPT Plus and Teams subscribers, alongside the cheaper o1-mini variant.
Contents4 sections
OpenAI o1 (ChatGPT o1)
OpenAI is pressing ahead with strong determination to build the world's first Artificial General Intelligence (AGI). It has rolled out its first step toward that goal, the new OpenAI o1 AI model, previously also known as Project Strawberry. However, since updates and improvements to the app are still ongoing, only a preview version is currently available. What is ChatGPT o1? Let's take a look at what we know about OpenAI o1 preview!
What Is OpenAI o1 (ChatGPT o1)?
During development, o1-preview was given a qualifying exam for the International Mathematical Olympiad; on a test where the previous GPT-4o model could only answer correctly 13% of the time, the new model answered 83% of the exam's questions correctly.
Also known as Project Strawberry, OpenAI o1 is OpenAI's latest large language model. It uses a combination of reinforcement learning and "chain of thought" reasoning. With its human-like reasoning abilities, it far surpasses the capabilities of current state-of-the-art systems. The model was said to power the next generation of ChatGPT. In fact, it's recently been confirmed that its success rate is far higher than the GPT-4o model's, and that it knows how to solve most of the math problems it's given. That said, it's fair to say the app still has a number of shortcomings. Let's take a more detailed look at the model's features:
What Are OpenAI o1's Features?
Besides math problems, OpenAI o1's known capabilities include being able to build marketing strategies and autonomously solve complex word puzzles like Connections. This new model can also autonomously browse the internet and even carry out in-depth research.
The model comes in two base variants: o1-preview and o1-mini.
o1-mini, the more cost-effective option, delivers strong performance at roughly 80% lower cost. It performs especially well on math and programming tasks and is ideal for applications that need fast solutions.
o1-preview can be used especially for solving multi-step problems, code debugging, and advanced scientific research, while o1-mini offers a good option for those looking for more cost-efficient AI solutions.
When Will OpenAI o1 (ChatGPT o1) Release?
In September 2024, OpenAI launched a preview of Project Strawberry called "o1-preview." It was accompanied by "o1-mini," a smaller and lighter version of the new model. At launch, the new model iteration was made immediately available to ChatGPT Plus and Teams subscribers. It's also known that the model will be rolled out to other users very soon.
And speaking of the AI detection tool GPTZero, how do you use GPTZero?
Why Was It Called Project Strawberry?
We don't know the exact reason behind the name "Strawberry," since OpenAI hasn't made that public. We believe it was likely chosen as an internal reference and a code name to maintain confidentiality during development.
About OpenAI
OpenAI is a US-based artificial intelligence research company. The company's overall goal is to conduct research into artificial intelligence that can benefit humanity. Founded in San Francisco in late 2015, the company aims to make its patents and research public, pursuing "open collaboration" with other institutions and researchers.
About Artificial Intelligence
Artificial intelligence, or AI for short, is the kind of intelligence displayed by machines, as opposed to natural intelligence displayed by animals, including humans. The distinction between the first and second categories is often reflected in the chosen abbreviation.
For the latest news, don't forget to follow the Gamer Papers website and our social media accounts.
Frequently asked questions
Questions answered in this article
What Is OpenAI o1 (ChatGPT o1)?
Also known as Project Strawberry, OpenAI o1 is OpenAI's latest large language model. It uses a combination of reinforcement learning and "chain of thought" reasoning. With its human-like reasoning abilities, it far surpasses the capabilities of current state-of-the-art systems. The model was said to power the next generation of ChatGPT. In fact, it's recently been confirmed that its success rate is far higher than the GPT-4o model's, and that it knows how to solve most of the math problems it's given. That said, it's fair to say the app still has a number of shortcomings. Let's take a more detailed look at the model's features:
Go to sectionWhat Are OpenAI o1's Features?
Besides math problems, OpenAI o1's known capabilities include being able to build marketing strategies and autonomously solve complex word puzzles like Connections. This new model can also autonomously browse the internet and even carry out in-depth research. The model comes in two base variants: o1-preview and o1-mini. o1-mini, the more cost-effective option, delivers strong performance at roughly 80% lower cost. It performs especially well on math and programming tasks and is ideal for applications that need fast solutions. o1-preview can be used especially for solving multi-step problems, code debugging, and advanced scientific research, while o1-mini offers a good option for those…
Go to sectionWhen Will OpenAI o1 (ChatGPT o1) Release?
In September 2024, OpenAI launched a preview of Project Strawberry called "o1-preview." It was accompanied by "o1-mini," a smaller and lighter version of the new model. At launch, the new model iteration was made immediately available to ChatGPT Plus and Teams subscribers. It's also known that the model will be rolled out to other users very soon. And speaking of the AI detection tool GPTZero, how do you use GPTZero?
Go to sectionWhy Was It Called Project Strawberry?
We don't know the exact reason behind the name "Strawberry," since OpenAI hasn't made that public. We believe it was likely chosen as an internal reference and a code name to maintain confidentiality during development. OpenAI is a US-based artificial intelligence research company. The company's overall goal is to conduct research into artificial intelligence that can benefit humanity. Founded in San Francisco in late 2015, the company aims to make its patents and research public, pursuing "open collaboration" with other institutions and researchers. Artificial intelligence, or AI for short, is the kind of intelligence displayed by machines, as opposed to natural intelligence displayed…
Go to sectionSources
External links cited in this article


