MediaTek announced on Weibo the day before yesterday that its Dimensity automotive cockpit platform C-X1 and Dimensity flagship mobile chips had achieved Day-0 adaptation for Alibaba’s Qwen 3.8 model.
This rapid adaptation means that smartphones and vehicles powered by Dimensity chips will be able to access the latest version of Alibaba’s Qwen large model immediately, allowing users to experience a smarter and more intuitive on-device AI experience in the future.
Alibaba open-sourced the Qwen3.8-27B model on August 14. All developers, research institutions, and enterprises can freely download, deploy, and use it.
It is understood that 27B (27 billion parameters) is the model size most highly anticipated by the global AI community. Qwen3.8-27B is a native multimodal Dense model that supports a 262K native context and can be extended to a 1M-token context through YaRN technology.
Compared with Qwen3.6-27B, the new model delivers substantially improved performance in programming and office scenarios, even surpassing Qwen3.7-Plus. It also adds the reasoning_effort feature, which controls the depth of reasoning based on task difficulty to reduce resource consumption.
Qwen3.8-27B also shares the common capabilities of the Qwen3.8 series, enabling it to complete complex tasks end to end and reportedly deliver reliable results that “stand up to scrutiny.”