It is also the most expensive open source frontier model, per task; cf. Cost per Intelligence Index Task. If it is as good as the benchmarks indicate -- I'll never know because I see no reason to try it -- it is a positive indicator for Qwen and China.