Arm Lumex Ushers in a New Era of On-Device AI, Delivering Critically important Performance Boosts
Breaking News: Arm today announced Lumex, its most advanced consumer system-on-chip (CSS) platform, designed to fundamentally shift AI processing to the edge – directly onto consumer devices.Lumex isn’t just an upgrade; it’s positioned as the foundation for a new generation of personal, private, and high-performance AI experiences, promising double-digit performance gains for key AI workloads.
The platform’s core innovation centers around deep integration with Scalable Matrix Extensions 2 (SME2), a technology enabling low-latency, quantized inference for large language models (LLMs). This allows complex AI models, previously relegated to the cloud, to run efficiently on smartphones and other consumer electronics.
“Arm Lumex is more than our most advanced CSS platform for the consumer computing market, it’s the foundation for the next era of bright AI-enabled experiences,” Arm stated in a press release. “Whether your an OEM or developer, Lumex gives you the tools to deliver personal, private and high-performance AI at the edge, where it matters most. Built for the AI era, Lumex is where the future of mobile innovation begins.”
Early validation results demonstrate the power of this approach. Collaboration between Arm, Alipay, and vivo has shown that LLM inference using SME2 on vivo’s next-generation flagship smartphone improves prefill performance by over 40% and decode performance by over 25%. “The validation of LLM inference using SME2 has been completed…We observe that prefill and decode performance can be improved by over 40% and 25% respectively,” confirmed Xindan Weng, Head of Client Engineering, Alipay.
The impact extends beyond performance metrics. Industry leaders are already embracing Lumex and SME2 to unlock new AI capabilities:
Alibaba: “Through deep integration with SME2, MNN enables low-latency, quantized inference for billion-parameter models like Qwen on smartphones – showcasing Arm and Alibaba’s joint innovation in scalable, next-gen mobile AI,” said Xiaotang Jiang, Head of MNN, Taobao and Tmall Group. Google: Iliyan Malchev, Distinguished Software Engineer, Android at Google, highlighted the broad applicability, stating, “SME2-enhanced hardware enables more advanced AI models, like Gemma 3, to run directly on a wide range of devices…it will enable mobile developers to seamlessly deploy the next generation of AI features across ecosystems.”
Honor: The company plans to leverage Lumex to deliver “smooth performance, intelligent AI features, and outstanding power efficiency” in its upper mid-range smartphones.
Meta: Sy Choudhury, Director, AI Partnerships, Meta, expressed excitement about the integration of arm Kleidi and PyTorch’s ExecuTorch, enabling seamless application performance on next-generation technology.
Samsung electronics: Nak Hee Seong, Vice President and Head of SOC IP Advancement Team, affirmed Samsung’s continued collaboration with Arm, aiming to “push the boundaries of on-device AI, delivering smarter, faster, and more efficient experiences.”
Tencent: Felix Yang, Distinguished Expert, Machine Learning Platform, Tencent, noted that SME2 accelerates on-device LLMs like Tencent’s Hunyuan, addressing performance bottlenecks and enhancing user experiences.
Arm Lumex represents a significant step towards a future where powerful AI is accessible and responsive, directly within the devices consumers use every day. The platform’s focus on edge computing promises not only faster performance but also enhanced privacy and reduced reliance on cloud connectivity.
