On the evening of August 14, Alibaba's QianWen team officially made the Qwen3.8 series models available as open-source. Notably, the Qwen3.8-27B model stands out as a native multimodal dense model, boasting an impressive 27 billion parameters. It surpasses the performance of its predecessor, Qwen3.7-Plus, by offering swift responses, high-quality output, and effortless deployment capabilities. Following quantization, this model can operate seamlessly on consumer-grade graphics cards. Moreover, it is released under the Apache 2.0 license, ensuring broad accessibility and usage rights.

Previously, the flagship model, Qwen3.8-Max, had already open-sourced its weights. As of now, Alibaba has contributed over 460 models to the global AI community. The Qwen models, in particular, have garnered immense popularity, amassing over 3 billion downloads worldwide and inspiring the creation of more than 300,000 derivative models. This makes the Qwen series the most downloaded open-source model on a global scale.