Arm Lumex Compute Subsystem Platform: Intelligence and Efficiency Redefined for On-Device AI
Arm Lumex Compute Subsystem (CSS) Platform is an advanced subsystem designed to bring AI-first experiences directly into the hands of consumers. It combines leading-edge new CPU clusters, a revolutionary new GPU, ultra-efficient system interconnect, and the latest Arm physical implementations, with seamless software integration, all designed to support heterogeneous AI compute at scale.
Arm Lumex CSS Platform redefines mobile performance, developer agility, and consumer experience by making AI pervasive and accessible, from flagship devices to cost-sensitive tiers.
Why Choose Arm Lumex CSS?
Advanced Subsystem for Flagship Mobile AI Performance
Arm Lumex CSS introduces a holistic re-architecture across CPU, GPU, and system IP, all purpose built to drive innovation in premium smartphones. Every subsystem is optimized for real-world AI workloads, from high-throughput compute to low-latency system responsiveness.
5x AI Performance Within the Same Power Budget
Thanks to Arm C1-Ultra with SME2 integration and architectural efficiencies, Arm Lumex CSS introduces 5x the AI performance uplift within the same power budget as previous generations, empowering developers to deploy larger, more responsive models without sacrificing battery life.1
2x Ray Tracing Performance for Mobile Gaming
The next-generation Mali G1-Ultra GPU doubles ray tracing performance over its predecessor, enabling photorealistic lighting, reflections, and immersive effects that redefine mobile gaming and XR quality.2
AI at Your Fingertips: Real-World On-Device Intelligence That’s Always Ready
Enhanced photography, immersive gaming, smart assistants, and real-time translation now run on device with low latency, privacy, and efficient power use.
Smart AI-Powered Photography
AI-powered vision technology that auto-applies creative filters while intelligently editing and correcting details, delivering the perfect shot every time.
Immersive Graphics for Mobile Gaming
Immersive, rich gaming is now possible on mobile devices thanks to the Mali G1 GPU with Arm Ray Tracing v2, delivering smooth, high-fidelity graphics with exceptional power efficiency. Built on Arm Lumex CSS, this experience brings optimized high-end visual performance to mobile devices, anytime, anywhere.
Context-Aware Personal AI Assistant
Make every smartphone run smoother with AI that proactively manages tasks, optimizes schedules, suggests faster routes, and surfaces timely app interactions. Running entirely on-device with minimal power usage, Arm Lumex CSS enables this intelligent, battery-efficient experience that feels like a personal assistant in the pocket.
On-Device Real-Time Voice Translation
Speak naturally, understand instantly. Arm Lumex CSS enables seamless, secure, real-time voice translation directly on the device, delivering fast, private, and always available communication, even without internet access.
Built for AI Performance, Efficiency, and Scale
Arm Lumex CSS boosts intelligent compute with on-device performance, efficiency, and developer-ready tools accelerating machine learning workloads, extending battery life, and streamlining implementation.
Arm Lumex CSS delivers up to 5x uplift in machine learning tasks, enabling on-device intelligence with low latency and real-time responsiveness.1
Designed with Arm SME2 and the latest efficient interconnect, Arm Lumex CSS platform provides up to 3x energy savings compared to previous generations, extending battery life even for intensive AI workloads.1
Arm provides ready-to-use physical designs to reduce development time and accelerate time to market for partners.
Developer readiness with tools for profiling, debugging, and optimizing AI applications across CPU, GPU, and system IP.
Core Features of Lumex CSS Platform
Built on Armv9.3-A with Arm SME2 for dedicated AI compute.
Includes the Arm Mali GPU with Ray Tracing v2 for AI-enhanced visuals.
SI L1 and MMU L1 System IP ensure best compute performance, longer battery life and security of CSS platform.
Provides faster time to market by enabling innovation at the speed and scale that AI demands.
AI readiness with unified software and developer tools.
Ecosystem and Partner Collaborations
Hear how innovative partners like Google, Meta, Samsung, Tencent GiiNEX, Honor, Alipay, and Alibaba are collaborating with Arm to accelerate on-device AI innovation.
“SME2-enhanced hardware enables more advanced AI models, like Gemma 3, to run directly on a wide range of devices. As SME2 continues to scale, it will enable mobile developers to seamlessly deploy the next generation of AI features across ecosystems. This will ultimately benefit end-users with low-latency experiences that are widely available on their smartphones.”
Iliyan Malchev
Distinguished Software Engineer, Android, Google
“AI is changing how we interact with our devices and the world around us, and the Arm ecosystem is driving important developments in this space. At Meta, we're excited about the integration of Arm Kleidi and PyTorch’s ExecuTorch mobile/edge runtime, allowing our applications to seamlessly run on next-generation technology that accelerates end-user experiences.”
Sy Choudhury
Director, AI Partnerships, Meta
“We’re excited to continue our collaboration with Arm by leveraging Arm’s Compute Subsystem platform to develop the next generation of flagship mobile products. This partnership enables us to push the boundaries of on-device AI, delivering smarter, faster, and more efficient experiences for our users.”
Nak Hee Seong
Vice President and Head of SOC IP Development Team at Samsung Electronics
“Arm Scalable Matrix Extension 2 (SME2) is a true game-changer for AI in gaming. By rethinking matrix multiplication at its core, SME2 delivers up to 2.5X performance gains and even greater acceleration under real-world loads. Tencent GiiNEX engine is delighted to make SME2 part of our solution, empowering developers to build richer worlds, smarter NPCs, and seamless real-time experiences that players will love."
Tencent - GiiNEX
“At Honor, our mission is to bring premium experiences to more users, especially through our upper mid-range smartphones. By leveraging the Arm Lumex CSS platform, we’re able to deliver smooth performance, intelligent AI features, and outstanding power efficiency that elevate everyday mobile experiences.”
Honor
“With Arm’s SME2, Alipay accelerates on-device large language models to deliver real-time, secure, and responsive AI experiences across everyday mobile scenarios.”
Xindan Weng
Head of Client Engineering, Alipay
“SME2 accelerates on-device large model inference, like Tencent’s Hunyuan, by addressing key performance bottlenecks and enabling efficient LLM deployment on mobile for enhanced user experiences.”
Felix Yang
Distinguished Expert, Machine Learning Platform, Tencent
“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.”
Xiaotang Jiang
Head of MNN, Taobao and Tmall Group, Alibaba
Build Next-Gen Mobile AI with Arm Lumex CSS
Access key tools and SDKs to optimize generative AI and image super-resolution on Arm-based devices.
Accelerate AI on Arm Devices with SME2
Explore developer resources for deploying on-device generative AI using Arm-optimized toolchains and models.
Master AI with SME2 on Arm CPUs
Dive into the SME2 Programmer’s Guide and learn how to leverage SME2 for high-throughput, low-latency AI acceleration—right on the CPU.
Optimize Code with Arm Performance Studio
Leverage powerful profiling tools like Streamline and Performance Advisor, optimized specifically for GPU & CPU architectures, to identify bottlenecks, boost FPS, reduce power consumption, and enhance user experiences.
Talk with an Expert
Learn how Arm Lumex CSS can help you create superior AI experiences on consumer devices.
Latest News and Resources
- NEWS and BLOGS
- Developer
Boost Android App Efficiency with SME2
Learn how SME2 helps accelerate AI performance on Arm CPUs, enabling more responsive and efficient mobile experiences.
Frequently Asked Questions: Understanding Arm Lumex CSS
Arm Lumex CSS is designed for AI-first smartphones. It integrates Armv9.3-A CPUs with SME2 for AI acceleration, a Mali GPU with Ray Tracing v2, and unified software tools to enable real-time, on-device intelligence.
Arm Lumex CSS Platform delivers up to 5x uplift in machine learning performance and 3x greater energy efficiency compared to previous generations, enabling always-on AI features, faster photo processing, and real-time user interactions.
- Armv9.3-A CPUs with SME2 for vectorized AI compute.
- Mali GPU with Ray Tracing v2 for high-end graphics and AI-enhanced visuals.
- New optimized System IP: Re-designed System Interconnect (SI) and optimized System Memory Management Unit (MMU).
- Kleidi AI Software for developer-ready AI workflows.
- Unified telemetry tools for real-time system insights.
- Ready-to-use physical implementations for reduced time to market.
Arm Lumex CSS powers a range of AI-first experiences in smartphones and mobile devices. Real-world use cases include:
- AI-enhanced photography: Automatically adjust lighting, remove blur, and enhance image quality in real time.
- On-device fitness coaching: Guide workouts and monitor performance with AI-driven motion tracking and analysis.
- Voice assistants and translators: Enable real-time translation and personalized voice interaction without relying on the cloud.
- Intelligent productivity tools: Support AI-generated content, music creation, and real-time text enhancements directly on the device.
These applications demonstrate how Arm Lumex CSS enables fast, private, and immersive AI features directly on consumer devices, no cloud dependency required.
Arm Lumex CSS introduces Armv9.3-A CPUs, SME2, and the Mali G1 GPUs with advanced ray tracing. These compute components are integrated with the new re-designed SI L1 system interconnect and optimized MMU L1 system memory management unit. It also features unified developer tools and ready-to-use implementations, dramatically reducing complexity and development cycles.
You can begin by accessing: