OpenAI, the pioneering company in the field of artificial intelligence, has announced its latest breakthrough: the models o3 and o3 Mini. These new solutions, specifically designed to tackle tasks requiring advanced reasoning, promise a significant leap forward compared to the previous generation of models such as o1. Although they are not yet available to the general public, they are already generating excitement in the technology community.
Since their initial release, reasoning models have proven to be key tools in applications that demand precision and detailed analysis. With o3 and o3 Mini, OpenAI seeks to reinforce its leadership in the sector, presenting substantial improvements in both performance and flexibility of use.
Highlights of the new models
- Models o3 and o3 Mini: Designed to deliver advanced reasoning and solve complex problems with greater accuracy.
- Improved performance: They surpass their predecessor o1 in capabilities, excelling in coding, mathematics and scientific reasoning benchmarks.
- Resource Optimization: o3 Mini is a lighter version, suitable for lower capacity devices, while o3 maximizes performance in advanced environments.
- Availability: Currently in testing for security researchers, o3 Mini is expected to be available in early 2025, followed by o3.
One of the highlights of o3 is its ability to solve complex problems by breaking them down into smaller steps. This process, known as the “chain of thought,” allows the model to analyze each stage before offering a final answer. This ensures not only greater accuracy, but also a detailed explanation of its reasoning, something invaluable for sectors such as education, advanced programming and scientific analysis.
The tests carried out so far have yielded impressive results. For example, in the benchmark SWE-Bench Verified, o3 achieved a 22,8% increase compared to its predecessor o1, reaching an accuracy level of 84,9%. In another notable test, the American Invitational Mathematics Exam 2024, o3 scored 96,7%, missing just one question.
In addition, the o3 Mini, an optimized and lighter version, is designed for specific applications and devices with less computational capacity. Both models benefit from a unique feature: users can adjust the «reasoning time» between low, medium and high. This allows for prioritising speed or accuracy depending on the context, a significant advance compared to previous models.
A strategic step towards general artificial intelligence
OpenAI has not overlooked the importance of branding and its impact on the market. According to the company, they decided to skip the “o2” name to avoid potential legal conflicts with British telecommunications company O2. This move, while curious, reflects the attention to detail of the firm led by Sam Altman.
Beyond the technical improvements, the new models also represent a crucial step towards artificial general intelligence (AGI). In tests such as the ARC-AGI, designed to measure human-like cognitive abilities, o3 scored 87,5%, far outperforming its predecessors.
Availability and Early Access
While the models aren’t ready for public release yet, OpenAI has opened a waiting list for security researchers interested in trying out the o3 Mini. This model will be available in early 2025, followed later that year by the o3. The company promises that both will be available on a subscription basis, though specific details have not yet been revealed.
This phased approach allows OpenAI to conduct extensive testing to ensure the safety and efficiency of its models, a critical aspect in developing advanced artificial intelligence.
A competitive market: OpenAI vs Google
The launch of the o3 and o3 Mini models comes at a time of intense competition in the AI sector. Google, for its part, has recently launched its “Thinking Mode” in Gemini 2.0 Flash, a technology that also seeks to improve logical reasoning in AI models. However, OpenAI seems to lean towards a more gradual and focused approach on specialization, unlike its competitor’s more versatile model.
Both companies are pushing the boundaries of what is possible in artificial intelligence, making it clear that the future of this field will depend on both technical advances and implementation strategies.
The o3 and o3 Mini models represent a significant advance in the search for more precise, reliable and specialized artificial intelligence. Their capabilities to tackle complex problems with structured logic open up new possibilities for scientific, professional and educational applications, marking a turning point in the development of advanced reasoning technologies.
I am a technology enthusiast who has turned his "geek" interests into a profession. I have spent more than 10 years of my life using cutting-edge technology and tinkering with all kinds of programs out of pure curiosity. Now I have specialized in computer technology and video games. This is because for more than 5 years I have been writing for various websites on technology and video games, creating articles that seek to give you the information you need in a language that is understandable to everyone.
If you have any questions, my knowledge ranges from everything related to the Windows operating system as well as Android for mobile phones. And my commitment is to you, I am always willing to spend a few minutes and help you resolve any questions you may have in this internet world.