MediaTek rapidly integrates Alibaba's Qwen 3.8 model into Dimensity platforms

MediaTek announced the immediate adaptation of Alibaba's Qwen 3.8 model for its platforms, including the Dimensity C-X1 automotive cabin solution and its Dimensity mobile chipsets. With this integration, smart devices and vehicles equipped with these processors can utilize the latest version of Alibaba's Qwen large language model from its launch, providing users with more advanced and intuitive on-device artificial intelligence experiences. Alibaba released the Qwen 3.8-27B model as open source on August 14, enabling developers, research institutions, and companies to freely download, deploy, and use the model. The 27 billion parameter size is considered the most popular by the global AI community. The Qwen 3.8-27B is a native dense and multimodal model that supports 262,000 context tokens and can expand to one million tokens through YaRN technology. Compared to the previous Qwen 3.6-27B version, the new model demonstrates significantly improved performance in programming and productivity scenarios, even surpassing Qwen 3.7-Plus. The model also introduces the reasoning_effort feature, which allows adjusting the depth of analysis based on task complexity while saving computational resources. Like other models in the Qwen 3.8 series, this model can execute complex tasks end-to-end, delivering reliable and verifiable results.