Microsoft Research and Huawei jointly announced a key commercial milestone for 1-bit large language models (LLMs), stating that BitNet b1.58 technology has realized the efficient deployment of large-scale models on edge devices equipped with Huawei's latest Kirin chips and Qualcomm's Snapdragon platforms. Supported by BitNet's innovative ternary weight architecture {-1, 0, 1}, the technology reduces the model's memory footprint by over 70% and consumes only 0.028 joules per inference, enabling large-scale LLMs to run on consumer mobile hardware without relying on cloud connectivity. This breakthrough optimizes the economics of on-device AI, making high-performance model capabilities accessible on smartphones and tablets without the need for data center infrastructure or network latency.

Kunlun Tech (昆仑万维) has officially launched Skywork Desktop (天工 Skywork 桌面版) globally on February 4, 2026, positioning it as a privacy-forward, on-device productivity agent that can read and organize large volumes of files directly on a user’s computer—without uploading documents to the cloud. Skywork Desktop emphasizes “content understanding over file formats,” supporting unified semantic processing across documents, spreadsheets, PPT, images, video, and more, while also enabling multi-task parallel execution for workflow automation and content generation

Apple is undertaking its most significant AI overhaul to date with the “Siri AI Rebuild Plan,” aiming to transform its longstanding voice assistant into a deeply integrated, context-aware, and on-device intelligence by leveraging a new large language model foundation and revamped system architecture, potentially announced at WWDC 2026.

Google has officially released TranslateGemma, a new suite of open-source machine translation models built upon the Gemma 3 architecture. Available in 4B, 12B, and 27B parameter sizes, these models deliver state-of-the-art translation quality for 55 language pairs while achieving a remarkable efficiency breakthrough. The 12B variant notably outperforms the baseline Gemma 3 27B model, offering high-fidelity translation with less than half the parameters