Khaberni - Chinese company Alibaba announced the development of a new artificial intelligence model named Qwen3-Max-Thinking, which achieved a perfect score in prominent global sports competitions.
Thus, it becomes the first Chinese model to achieve this feat and directly competes with advanced American AI models like GPT-5 Pro from "OpenAI".
The company stated that the new model scored 100% accuracy in both the American Mathematics Competition (AIME 2025) and the Harvard-Massachusetts Mathematics Tournament (HMMT), which are among the world’s most complex contests in the fields of arithmetic, algebra, number theory, and probability, according to a report published by "scmp".
Intuition Labs, a company specializing in AI software, said that the performance of models in these competitions is a pivotal test for their ability to think logically and solve problems, noting that the race of mathematical models has become one of the most prominent areas of competition among technology giants.
The new model is based on the Qwen3-Max platform, which has more than a trillion coefficients, and was first introduced in September last year.
Alibaba Cloud confirmed that the performance of the model equals or exceeds global competitors such as Claude Opus 4 from "Anthropic" and V3.1 from "DeepCic" and Grok 4 from "xAI", alongside GPT-5 Pro from "OpenAI".
Qwen3-Max recently won in a real-world test of cryptocurrency trading against five other American and Chinese models, achieving a return of 22.3% on an initial investment of $10,000 over two weeks, while American models recorded losses, with GPT-5 performing the worst with a loss reaching 62.7%.
The model is now available for individual users through the Qwen Chat platform or via an API interface provided by Alibaba Cloud, while researcher Lin Junyang from the development team confirmed that work is still ongoing to improve the model's capabilities, stating: "The task is not finished yet."
