OpenAI, the company behind ChatGPT, has unveiled o1, a new artificial intelligence model designed to excel at complex reasoning, advanced mathematics, and programming tasks. The launch marks a paradigm shift in how AI systems approach problem-solving. 

Unlike traditional models that rely on statistical predictions to generate answers, o1 introduces a process that mirrors trial-and-error reasoning. Instead of rushing to provide the most “probable” response, it takes additional time to analyze variables, construct logical chains, and validate its own outputs. 

Early benchmarks suggest the leap is substantial. On an International Mathematical Olympiad qualifying exam, GPT-4o scored correctly on only 13% of problems, while o1 achieved an impressive 83%. In coding competitions hosted on Codeforces, the new model ranked in the 89th percentile, rivaling skilled human developers. 

The model also shows potential across high-level scientific fields. Health researchers could employ it to annotate cellular sequencing data, while physicists might use it to generate advanced equations for quantum optics. In software development, o1’s ability to execute multi-step workflows introduces new efficiencies for teams tackling complex projects. 

Limitations of o1

Still, o1 has limitations. Unlike GPT-4o, it cannot generate images or browse the web in real time. While it hallucinates less frequently, it can still produce inaccurate content. OpenAI CEO Sam Altman has cautioned that the model “remains limited and often feels more impressive on first use than after extended interaction.” 

OpenAI will release two versions: o1 and o1-mini, the latter being a lighter, more affordable alternative. ChatGPT Plus and Team users already have access, while Enterprise and Edu customers will gain access next week. Eventually, free-tier ChatGPT users will also be able to try out the system. 

Pricing reflects the system’s complexity. API access to o1 costs $15 per million tokens processed and $60 per million tokens generated, compared with GPT-4o’s $5 and $15 rates. This highlights both the higher computational demands and the premium placed on advanced reasoning capabilities. 

The announcement comes as OpenAI seeks a new $6.5 billion funding round that could push its valuation to $150 billion. The move underscores how rapidly the artificial intelligence market is evolving and how investors are betting heavily on technologies that could redefine science, business, and daily life. 

With o1, OpenAI aims not just to surpass ChatGPT, but to pioneer a new generation of AI—one that doesn’t merely answer, but thinks.