The latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.