Alibaba AI aces math tests usually meant for students
Alibaba's new AI, Qwen3-Max-Thinking, nailed perfect scores at both the AIME and the AMC—two competitions usually reserved for top student mathletes.
This is a first for any Chinese AI, and it puts Alibaba in direct competition with US tech giants when it comes to advanced reasoning.
Qwen3-Max-Thinking didn't just spit out answers
These aren't your average tests—they're packed with tricky algebra and number theory questions that stump even brilliant students.
What's cool is Qwen3-Max-Thinking didn't just spit out answers; it showed its work step by step, proving it can actually "think" through problems like a human.
Qwen3-Max-Thinking runs on Alibaba's trillion-parameter model
Qwen3-Max-Thinking runs on Alibaba's trillion-parameter model and uses a "Thinking Mode" to break down tough problems into smaller chunks—kind of like how you'd tackle a big homework assignment.
With access through Alibaba Cloud APIs and supported platforms, Alibaba is hoping to shake up not just math contests but also real-world fields like science, engineering, and finance.