OpenAI o1, the artificial intelligence at a new level of reasoning

The modern era of artificial intelligence began with the development of large language models, such as ChatGPT (2023), and the initial development of key architectures and algorithms, such as the transformer architecture, in 2017, which led to the scaling and development of models languages ​​that exhibit human-like traits of reasoning, cognition, attention, and creativity. In 2020, the GPT-3 model by OpenAI was released. In 2023 Microsoft Research tested the GPT-4 large language model on a wide variety of tasks and concluded that it could reasonably be viewed as an early (but still incomplete) version of a general artificial intelligence system. OpenAI is currently developing ChatGPT 5 which will be a next-generation language model, the fifth iteration of the Generative Pre-training Transformer (GPT) language model, a massive leap forward in the field of natural language processing.

On September 13th, OpenAI took a leap by announcing a new “o1” family of AI models designed to reason through complex tasks and solve harder problems than previous models in the GPT series. The new series started with two versions, “o1-preview” and “o1-mini”, both of which were initially limited to 30 messages per week. According to OpenAI, “As an early model, it doesn’t yet have many of the features that make ChatGPT useful, like browsing the web for information and uploading files and images. For many common cases GPT-4o will be more capable in the near term.”

OpenAI claims that its new o1 series is particularly useful for researchers in solving more complex problems, including generating complex mathematical formulas. Third-party tests have shown that o1-preview performs at a level close to that of PhD students in fields such as physics, chemistry and biology. o1-mini is efficient for building and executing multi-step workflows, debugging code, and efficiently solving programming challenges.

o1-mini will soon be offered to all users with a free account.

OpenAI plans to regularly update and improve these models, including adding features such as browsing, uploading files and images, and calling functions, which are currently not available in the API version, in parallel with the development of the GPT series.

I asked ChatGPT and o1 what they can tell me about this release.

ChatGPT 4.0 cites improved multimodal capabilities, increased efficiency, and deeper ethics and security as features of this series, saying it will revolutionize industries such as healthcare, education, entertainment, and customer service. “OpenAI O1 is the culmination of years of research and innovation. We’re excited to see how this model will transform the way people work, learn, and create,” said OpenAI CEO Sam Altman during the launch event.

o1-mini echoes OpenAI’s press release. “With O1, we’re pushing the boundaries of what’s possible in AI,” said Dr. Emily Chen, Chief Technology Officer at OpenAI. “This platform not only improves on existing technologies but also opens up new avenues for innovation across various sectors, from healthcare and finance to education and entertainment.” Key features and capabilities listed by o1-mini include improved language understanding, adaptive learning, scalability and integration, and robust security measures.

o1-preview seems more unprepared to answer questions about more recent events: “As of October 2023, there are no public records of OpenAI launching a product or service called “OpenAI O1.” It’s possible that this information emerged after my last update or hasn’t been widely publicized yet. I recommend checking OpenAI’s official website or recent news outlets for the most up-to-date information on their latest developments”.