OpenAI's AI System Solves All Tasks at ICPC World Finals 2025
OpenAI has made a groundbreaking achievement in artificial intelligence. Its AI system, a combination of GPT-5 and an experimental gemini model, solved all tasks at the International Collegiate Programming Contest World Finals 2025. This success places it ahead of human participants and Google's DeepMind competitor.
The OpenAI AI system demonstrated remarkable performance. It solved 11 out of 12 tasks using GPT-5, with the experimental gemini model stepping in to tackle the most challenging problem. This resulted in a perfect score of 12 out of 12, which would have placed it first among human participants.
This success is not a one-off achievement. The same models have consistently shown excellence, having previously secured gold-level results at the IMO and IOI. OpenAI highlights that this success is due to a more intelligent model with 'deeper thinking and longer computation times' in complex tasks.
The AI system competed under the same conditions as human participants. It received tasks in the same PDF format and had a five-hour time limit. For the last task, it required nine submissions, while the best human team achieved 11 out of 12.
OpenAI's AI system has set a new benchmark in AI capabilities. Its success at the ICPC World Finals 2025 is a testament to the potential of more intelligent gemini models in complex tasks. According to Mostafa Rohaninejad, involved in the success, the 'true milestone' for AI is discovering new knowledge. The 'general applicability' of these systems is also emphasized.
Read also:
- Hematology specialist and anemia treatment: The role of a hematologist in managing anemia conditions
- Trump announces Chinese leader's confirmation of TikTok agreement
- U.S. Army Europe & Africa Bolsters NATO, African Partnerships in Phase Zero
- Hackers Utilize GOLD SALEM to Infiltrate Networks and Evade Security Measures, Deploying Warlock Ransomware
