Recently, China's tech giant Alibaba Group released an upgraded artificial intelligence reasoning model - Qwen3-Max-Thinking. The model achieved a perfect score in two top global math competitions, demonstrating its strong capabilities in reasoning and problem-solving. This makes Qwen3-Max-Thinking the first Chinese AI reasoning model to achieve 100% accuracy in the American Invitational Mathematics Examination (AIME) and the Harvard-MIT Mathematics Tournament (HMMT).
Image source note: Image generated by AI
AIME and HMMT are among the most challenging math competitions globally, testing participants' problem-solving abilities in arithmetic, algebra, number theory, and probability. According to Intuition Labs, an AI software company based in San Jose, high scores in math reasoning tests are crucial for evaluating an AI model's reasoning and problem-solving skills. These competitions represent an important frontier in AI progress, highlighting the competition to develop stronger reasoning models.
The Qwen3-Max-Thinking model is the latest version in the Qwen3-Max series launched by Alibaba's AI and Cloud Computing division, featuring over 1 trillion parameters. Qwen3-Max was released at the end of September, while the original Qwen3 was launched in April. During the launch, Alibaba Cloud stated that Qwen3-Max stands out among domestic and international competitors, including Anthropic's Claude Opus4, DeepSeek's V3.1, xAI's Grok4, and OpenAI's GPT-5Pro.
Notably, Qwen3-Max also performed well in an experiment involving real markets, competing against five leading AI systems from the United States and China in cryptocurrency trading. Within two weeks, Qwen3-Max achieved a 22.3% return on investment, while other models performed poorly, with OpenAI's GPT-5 suffering a loss of 62.7%.
Currently, Qwen3-Max-Thinking is available to individual users through the web version of Qwen chatbot and Alibaba Cloud's API platform. Lin Junyang, a researcher from the Qwen team, revealed on social media that they are still continuing to improve this new reasoning model, stating that the work is not yet complete.
Key points:
- 📊 Alibaba's Qwen3-Max-Thinking achieved a perfect score in international math competitions, challenging OpenAI.
- 🏆 The model has over 1 trillion parameters and is the first Chinese AI to achieve 100% accuracy in AIME and HMMT.
- 📈 Qwen3-Max performed well in real-world cryptocurrency trading, achieving a 22.3% return on investment.
