Tag
Open source AI inference reaches 300 tok/s on mobile, with a WebGPU framework pushing Liquid AI's LFM2.5 230M to 1,400 tok/s in browser.
swSIM is an open-source, software-only SIM card simulator that works without any hardware and can connect to phones via a PC/SC interface.
Google has released its first standalone Google Finance app for Android, with an iOS version promised for later in 2026. The app includes AI-powered features like key moments for chart explanations and an AI research tool, alongside the general overhaul of the Finance website leaving beta.
Phil Schmid announces the launch of computer use in Gemini 3.5 Flash, enabling control of browser, mobile, and desktop environments, with a quickstart guide for controlling Android phones via adb.
Blåmba is an open-source re-implementation of early 2000s ringtone and multimedia providers for cellphones, supporting modern web and WAP access. It also includes a talk from the 38th Chaos Communication Congress covering the technical background.
Google begins rolling out Android 17 to Pixel phones and watches, introducing features like multitasking bubbles, foldable gaming mode (coming later), and native screen reaction video support.
Google rolls out Android 17 to Pixel phones, introducing Bubbles for multitasking, Screen Reactions for recording selfie videos during screen recording, and other features, with additional AI capabilities like Gemini Omni and Lyria 3 arriving later.
Apple's upcoming iOS 27 update promises significant performance improvements for older iPhones, including faster app launches, AirDrop transfers, and photo library loading, thanks to CPU scheduler optimizations and other platform enhancements.
Compares two locally running mobile TTS models, Kokoro and Supertonic, questioning their production quality beyond initial demos.
Proxee syncs your localhost to your phone, allowing you to access your local development environment on mobile.
ChatGPT now supports generating interactive charts directly from data and comparison results, available on both web and mobile platforms.
New QAT Gemma 4 checkpoints offer similar performance with ~4x less memory, enabling a 1GB memory footprint for Gemma 4 E2B via a new mobile quantization format.
Google released quantization-aware trained Gemma 4 checkpoints on HuggingFace, optimized for mobile device inference and available in QAT Mobile and Q4_0 variants.
An opinion piece argues that Google's Gemma4-12B model is strategically designed for IoT and mobile devices within the Android ecosystem, not just laptops as marketed, prioritizing low-latency speech and video processing over quality.
LM Studio enables running local open-source AI models on your phone by borrowing compute from your MacBook, 100% free.
A tool or app that enables users to run large local AI models directly from their iPhone, bringing on-device LLM inference to iOS.
An open-source Android launcher, GHOST, is nearing full release after a year of development, with implementation details available on GitHub.
TGIFolo releases version 1.9.0 for desktop and 0.5.4 for mobile, featuring enhanced AI summaries powered by Kimi, improved scroll mark-read with batching and retries, and cleaner Obsidian handling.
Xiaomi announces a refined design for the Xiaomi 17T Pro, aiming for a balance of screen visibility and hand feel.