As the curtains closed on the 2025 National College Entrance Examination, a parallel competition unfolded, testing the mettle of AI in tackling high-stakes math problems. In a rigorous assessment by professional media, an AI named Quark outperformed its peers, scoring 145 and 146 in two rounds, surpassing DouBao and setting a new benchmark for domestic AI in mathematical prowess.

The public evaluation saw models like Quark, DouBao, YuanBao, and ChatGPT compete using the 2025 national math exam, with internet search capabilities disabled, relying solely on deep contemplation mode for problem-solving. Quark emerged victorious in the assessment by Blue Whale Finance, securing the top spot with a score of 145. In multiple-choice and fill-in-the-blank questions, Quark achieved a 93% accuracy rate, leading the pack. However, all AI contestants stumbled on the sixth multiple-choice question, with analysis indicating that the error stemmed from AI's misinterpretation of vector coordinates and arrow directions in diagrams.

In the "Four Woods Theory" evaluation, Quark once again took the lead with a score of 146. In terms of speed, Quark was the fastest, followed closely by DouBao. For example, Quark could solve a problem within four minutes, while other products took an average of six minutes.

Behind Quark's problem-solving prowess lies the "Quark Learning Lingzhi Model." This model is based on the general understanding of a thousand questions, supported by a vast library of learning materials and post-training capabilities. It excels particularly in addressing complex problems in the sciences, offering users a new, heuristic learning experience.