I stumbled upon this story while scrolling through my feed, and honestly, I’m shocked. Some human still managed to beat AI in a head-to-head coding showdown. Przemysław Dębiak, a former OpenAI employee, barely alive after a punishing contest, claimed victory over one of OpenAI’s own models. It’s like watching a gladiator match where the human’s still standing, but just barely.
The contest was the AtCoder World Tour Finals 2025 Heuristic contest, held in Tokyo last week. Dębiak, competing under the name “Psyho,” faced off against multiple human coders and OpenAI’s OpenAIAHC, a simulated reasoning AI model related to OpenAI’s o3. The challenge was to solve a complex optimization problem over ten grueling hours, with scores based on the quality of solutions.
Humanity has prevailed (for now!)
I'm completely exhausted. I figured, I had 10h of sleep in the last 3 days and I'm barely alive.
I'll post more about the contest when I get some rest.
(To be clear, those are provisional results, but my lead should be big enough) pic.twitter.com/fIMo0ifNCd
— Psyho (@FakePsyho) July 16, 2025
Contestants could use any programming language allowed by AtCoder, but had to work on identical hardware and wait five minutes between submissions. Dębiak’s final score was 1,812,272,588,909, topping the leaderboard and beating the AI’s 1,654,675,725,406, a pretty tight race considering the stakes.
He posted his victory on X (formerly Twitter) with the defiant statement, “Humanity has prevailed (for now!)” yet confessed to being utterly drained after the marathon session. Ten hours of sleep over three days will do that to you.
OpenAI, in contrast, seemed content with its model’s silver medal. A company rep told Ars Technica that models like o3 rank in the top 100 for coding and math contests, but this was their first time snagging a top-three spot in a major competition. They praised the contest for testing strategic reasoning, long-term planning, and iterative improvement—just like a human would.
Worth notin,g though, Google DeepMind’s AI systems, AlphaProof and AlphaGeometry 2, scored silver medal level in a tough math competition last year, though they didn’t officially enter the event. This AtCoder contest is believed to be the first where an AI directly competed against human programmers.
While Dębiak’s win is a win for humanity, the fact that the AI came so close is a little unsettling if you ask me. Coders might be sharpening their skills, but AI development speed could soon tip the scales in favor of machines topping these leaderboards.
For now, though, Dębiak’s got the bragging rights, probably fueled by an energy drink and a much-needed nap. Chalk one up for the humans, but just barely.
