Asia Tech Wire (Aug 9) -- Chinese tech giant Alibaba has released Qwen2-Math, a family of next-generation math AI models that can beat GPT-4o and Claude-3.5.

Qwen2-Math is a series of language models built on Qwen2 LLM specifically for mathematical problem solving, ranking first globally in mathematical reasoning ability.

Evaluations across a range of math benchmarks show that the flagship model, Qwen2-Math-72B-Instruct, outperforms state-of-the-art models including GPT-4o, Claude-3.5-Sonnet, Gemini-1.5-Pro, and Llama-3.1-405B.

This flagship model handles a wide range of math problems in algebra, geometry, counting and probability, and number theory with 84% accuracy.