Chen joined the company in 2018 after working as a quantitative merchant at the signature of Wall Street Jane Street Capital, where he developed automatic learning models for futures trade. In Openai he headed the creation of Dall-E, the generative image image model. He then worked to add the recognition of images to GPT – 4 and led the development of Codex, the generative coding model that feeds the Github co -pilot.
Pachocki left an academic career in theoretical computer science to join Openai in 2017 and replaced Sutskever as a chief scientist in 2024. Together with Sutskever, he is one of the key architects of the so -called OpenAi reasoning models, especially O1 and O3, which are designed to address complex tasks in science, mathematics and codifications.
When we met, they were buzzing, just out of two new consecutive victories for their company’s technology.
On July 16, one of Openai’s great language models was secondly at the Atcoder World Tour finals, one of the most hardcore programming competitions in the world. On July 19, Openai announced that one of his models had achieved gold level results in the Mathematics Olympiad of 2025, one of the most prestigious mathematical competitions in the world.
The mathematical result reached the headlines, not only for the remarkable achievement of Openai, but because the rival Google Deepmind revealed two days after one of its models had achieved the same score in the same competition. Google Deepmind had played with the rules of the competition and waited for the organizers to verify their results before making an ad; Operai had indeed marked his own answers.
For Chen and Pachocki, the result speaks for himself. Anyway, it is the victory of programming that get most excited. “I think it’s quite underestimated,” Cen told me. A result of the gold medal at the Mathematics International Olympiad places you somewhere among the 20 and 50 competitors, he said. But in the Atcoder Contest Openai model placed in the first two: “Enter a really different level of human performance, which is not precedents.”
Boat, ship!
Operai people still like to say that they work in a research laboratory. But the company is very different from what was before the launch of Chatgpt three years ago. The company is now in a race with the largest and most rich technology companies in the world and valued at $ 300 billion. The investigation that pushes the containers and striking demonstrations no longer cut it. He needs to send products and put them in the hands of people, and boy, he does.
Operai has maintained an execution of new releases: obtaining important updates in its GPT-4 series, launching a chain of generative image and video models, and introducing the ability to speak with Chatgpt with your voice. Six months ago he started a new wave of so -called reasoning models with its O1 launch, soon followed by O3. And last week he launched the public’s user agent operator. Now he states that more than 400 million people use their products every week and have 2.5 billion indications per day.
#people #shape #Openais #future
