Amid the ongoing debate on AI vs. humans at work, a new example has been added to the favor of humans. A Polish Programmer, Przemysaw Debiak Aka Psyho, conquered an advanced AI model from Openai in a head-to-head programming battle at the iconic Atkodar World Tour Final in Tokyo at Tokyo. Despite the tireless efficiency of artificial intelligence, it was human determination that came to the top, if only by a narrow margin.

The competition organized by Japanese competitive programming platform Atkodar is widely considered one of the most difficult and most exclusive events in the region. Each year, only 12 of the world’s top-ranked programmers, the 2025 version introduced the first-its-way challenge: a straight face-off between human intelligence and a custom-made AI model from OpenAII, the company behind the chatgipt.

Although Openai sponsored the event and entered his model in a special “Human vs AI” exhibition match, it was his former employee, Debiak, who managed to exclude the machine. He had already competed in many incidents in the last three days, pushed himself to the limit. “I’m completely tired. … I am barely alive,” he posted on the social media platform X (East Twitter) after the match. “Humanity has become strong (for now!).”

His final score? A surprising 1,812,272,558,909 points. Openai’s model, listed under the name “Openahac”, finished second with 1,654,675,725,406 points, just 9.5 percent behind. AI still managed to improve the remaining ten elite human programmers, who made each a year -long competition through a year -long ranking.

A test of endurance

The format of the competition was fierce. Participants had a single complex adaptation problem under the Heuristic Contest Division to deal with 600 minutes, a full 10 hours, a single complex adaptation problem. These problems fall under the NP-Hard category, which requires the clever, often incomplete strategies to reach the best possible solutions within the lack of strict time.

All contestants, humans and AIs were provided with similar hardware to ensure fairness. Submission was allowed in any programming language supported by Atkodar, with a five -minute cold period between each.

In front of rapid automation, Debiyak’s victory for the flexibility of human intelligence has become a modern time metaphor.

The competition not only tested the raw programming ability, but also became a symbolic moment in the ongoing debate about the growing role of AI in skilled businesses.

Openai CEO Sam Altman Comments

The symbolic nature of the match was not lost on Openai. In a post on X, the company wrote: “Our model finished second in the Atkodar Hurstic World Final! Congratulations to the Champion for closing us at this time.”

Openai CEO Sam Altman also accepted Debiaq’s victory, posting a simple but message: “Good job Psyho.”

For Debiyak, it was a proud but tedious win, made his history even more important with OpenaiI, where he had worked earlier. The result acts as a reminder, while AI keeps increasing in capacity, human simplicity, under pressure, still keeps its land.

However, Openai called in second place, a milestone for AI in competitive programming ended, which raises widespread questions about the future of such competitions. If the AI models can already rival world -class humans under fair circumstances, how long will it be completely before dominating?

For now, the balance is maintained. But as -AI devices become sharp, clever and more autonomous, competitions such as Atkodar can become a fast battleground, not only for high scores – but for the soul of programming.