Alibaba Qwen · Qwen2
Alibaba's 72B Qwen2 flagship instruct model — top open-source performance on coding, math, and multilingual tasks at the time of release.
stce agent (Qwen2 family)
rank #58, IFEval=79.89 BBH=57.48 MATH=41.77 GPQA=16.33 MUSR=17.17 MMLU-Pro=48.92