Arm's October roundup: robocars, smart glasses, tablets and on-device AI models
Arm used its October 2026 ecosystem roundup to highlight partner products across autonomous vehicles, robotics, wearables, tablets and on-device AI models. The common thread, according to Arm, is a shared compute foundation that lets partners bring more intelligence into devices, infrastructure and the physical world.
- The Tensor RoboCar integrates 433 Arm-based cores for Level 4 autonomy, according to Arm.
- Alibaba's Qwen Glasses N1 and Amazon's Alexa Tablet 12 Pro are among the Arm-based consumer devices highlighted.
- Early benchmarks of Liquid AI's LFM3-1.7B on Arm CPUs show 3.1x faster prefill and 1.5x faster decoding with SME2.
- Arm AI Model Optimizer delivered up to 5.9x faster inference and 74% lower memory use for one model on Raspberry Pi 5.
- Google's Gemini Robotics Embodied Reasoning 2 is in public preview on the Gemini platform.
Autonomous mobility and robotics
The Tensor RoboCar integrates 433 Arm-based cores across the vehicle, according to Arm. It uses technologies including Arm Neoverse AE and Cortex processors to support centralized AI, agentic in-cabin experiences, distributed sensing and vehicle control. Arm describes the car as a showcase of scalable, power-efficient AI for Level 4 autonomy and has published a video on how its technology is distributed through the vehicle.
In prosthetics, PSYONIC's Ability Hand is a touch-sensitive prosthesis powered by 12 Arm-based microcontrollers that coordinate sensing, motor control and feedback. A BBC Tech Now report filmed at Arm's Cambridge headquarters featured PSYONIC founder and CEO Dr. Aadeel Akhtar and Drew Henry, EVP for Physical AI at Arm. The same hand can be used on robotic platforms, which opens the possibility of using human-generated grip and contact data to help train machines for more precise manipulation.
Agile Robots unveiled Agile ONE SN, a humanoid reference platform for researchers and developers. It brings NVIDIA Isaac GR00T and Jetson Thor compute, built on Arm Neoverse V3AE CPUs, into a compact research system. Arm says it aims to reduce the integration work that slows robotics research by connecting data capture, simulation, training and testing.
Shared frameworks and reasoning models for physical AI
Arm also pointed to the Robotics Capability Framework, which is meant to establish a common language for defining robotic capabilities and system requirements. According to Arm, this makes it easier to compare systems, reduce fragmentation and align development efforts.
On the model side, Google's Gemini Robotics Embodied Reasoning 2 is now available in public preview on the Gemini platform. It brings spatial reasoning, multi-step planning and multi-robot coordination into enterprise development workflows. Arm says it is working with Google to explore how these reasoning capabilities could shape future robotics compute platforms.
Consumer devices: glasses, tablets and PCs
Alibaba launched its Qwen Glasses N1 series, which Arm says features 50MP cameras, 4K video recording, AI assistance and hands-free payments. Himax's WiseEye processor, based on Arm Cortex-M55 and Ethos-U55, enables low-power, always-on visual sensing. By detecting relevant events locally and waking the main system only when needed, the design addresses the battery limits of intelligent eyewear.
Amazon introduced its first lineup of Alexa tablets, bringing Alexa+ to a wider range of everyday computing. The flagship Alexa Tablet 12 Pro uses a MediaTek Dimensity 8400 processor with Arm Cortex-A725 CPUs and Arm Mali-G720 GPU.
For PCs, Christopher Bergey, Executive Vice President of Edge AI at Arm, wrote a LinkedIn article titled "The next PC transformation: When computers adapt to us". It explores how agentic AI could make PCs more intuitive and able to act on users' behalf.
On-device models and optimization tools
Liquid AI launched its LFM3 family of compact, open-weight models for text, vision, audio and decision-making, with Arm supporting efficient inference from launch. Early, provisional benchmarks of the quantized LFM3-1.7B model on Arm CPUs show 3.1x faster prefill and 1.5x faster decoding with Arm Scalable Matrix Extension 2 (SME2) enabled, using just two CPU threads. Liquid AI is also among the partners in Arm Total Design for Physical AI.
Arm also highlighted its AI Model Optimizer, built on Nota AI's NetsPresso, which helps teams optimize, evaluate and compare models through Arm AI Portal. In early tests on a Raspberry Pi 5, one image-classification model ran up to 5.9x faster with 74% lower memory use.
What it means
The roundup shows how Arm positions its platform across very different products, from vehicles and prosthetics to glasses, tablets and small language models. For developers, the items point to better tooling for running and comparing models on real devices, and to shared standards in robotics. The performance figures for LFM3 are described as early and provisional, and all results come from Arm's own account.
