Arm Blog

Use the Arm AI Model Optimizer to optimize, evaluate, and compare your models so you can build your best AI applications on Arm

Learn how ExecuTorch brings PyTorch deployment to Arm Cortex-M CPUs with CMSIS-NN acceleration, low memory use, and support for embedded ML workflow today

Co-authored by Jason Zhu, Jackie Lee, and Damien Dooley (Arm), and Greyson Xu, Neo Zhang, and Yiming Li (Tencent Media Lab)

Explore how Arm adapters for Google Model Explorer help developers visualize AI models, inspect graph transformations, and identify performance bottlenecks

See how Arm engineers optimize streaming ASR with ExecuTorch and SME2, comparing model precision, word error rates and real-time performance on a Pixel 11

Explore how CMSIS-Ethos-U, the SDS Framework, and CMSIS Developer Assistant simplify Edge AI deployment, data replay, integration and debugging on devices

Explore how ExecuTorch, Zephyr and Arm Ethos-U55 bring PyTorch models to life on the Renesas RA8P1 through a practical camera-based edge AI application

Python 3.15 strengthens native development on Windows on Arm with broader package support, improved tooling, and native GitHub Actions workflows for teams

Celebrate 25 years of Buildroot and see how it helps Arm developers create reproducible embedded Linux systems for devices and intelligent edge workloads

Expanding Arm Learning Paths for developers and the AI agents they work with

Discover how the OCP RAS API standardizes fault management interfaces to reduce integration complexity across heterogeneous data center environments today

Learn how Arm and Google optimize ML Drift for Arm GPUs, improving on-device AI inference performance through kernel, resource, and workgroup tuning

Learn how evidence-first harness engineering combines indexed workload data, bounded LLM analysis, and simulation evidence to improve optimization results

Learn how Vulkan device-fault and address-binding extensions help diagnose VK_ERROR_DEVICE_LOST on Arm Mali GPUs and trace faults to app resources quickly

Discover how the 2026 Arm A-Profile architecture updates improve efficiency, software intent, memory management, and support for evolving AI workloads now

Discover Arm contributions to LLVM 23, including architecture support, performance improvements, code generation, tooling, libraries, and MLIR updates

See how LiteRT, quantization, KleidiAI, and SME2 improve on-device toxic speech detection on mobile, reducing latency and server processing demand overall

Learn how AMBA C2C extends chiplet connectivity across AXI and CHI with scalable protocol subsets, interoperability, and flexible transport support today

Learn how Arm Neural Frame Rate Upscaling works with Neural Super Sampling to improve motion, reduce rendering work, and support Unreal Engine games today

This blog post provides a brief overview of the MRS as discussed in a white paper. The white paper provides an introduction to the data structures, and potential use cases for working with the MRS dat…

Discover the winners of the Arm Create: AI Optimization Challenge and how their projects explore AI optimization across Cloud, Mobile, and Physical AI

Explore how ExecuTorch can partition and optimize one PyTorch model across Arm CPUs, GPUs, and NPUs using five target-specific execution paths

New Mali timeline and Dispatch graph views connect Vulkan submission, NX execution, shader activity, and system performance in one analysis workflow

How to find, optimize, and deploy AI models on Arm with pre-optimized models, performance insights, and agent-ready development workflows

Explore how Llama-Mobile uses S3D8 and quantization-aware training to reduce model size while retaining accuracy for multimodal AI on Arm mobile CPUs

Learn how to find and fix hidden graphics performance problems in mobile games using Arm Performance Studio.

Moving beyond a local-model demo requires more than an inference endpoint. Developers need an operational local-first runtime with explicit data boundaries