AI Briefing#on-device-ai
The latest AI and developer news about #on-device-ai, with the original source and a short summary.
The latest AI and developer news about #on-device-ai, with the original source and a short summary.
The model retains 98.2% of FP16 intelligence while reducing size to ~5.9 GB for llama.cpp.
The new Apache 2.0 licensed engine abstracts OpenGL ES, OpenCL, Metal, and WebGPU to enable real-time ML experiences across platforms.
The iPod creator argues that successful AI agents must operate on-device to earn consumer trust, citing Rabbit R1 and Humane Ai Pin as cautionary tales.
The 740M-parameter model unifies text, images, audio, and video into a shared embedding space under an Apache 2.0 license.
LiquidAI has released LFM2.5-Encoder-350M and LFM2.5-Encoder-230M, with the 350M variant described as a multilingual bidirectional encoder supporting 15 languages and on-device inference.
The model achieves an 11.1 ms time-to-first-token on an Apple M4 Pro CPU and supports seven languages.
The method achieves 2.8× compression while staying within 1.9% relative WER of the teacher model on five of six pairs.
Cactus Compute has released 'Needle 3', an ultra-compact foundation model ranging from 29MB to 121MB, stating that its 4-layer subnetwork achieves automation performance equivalent to DeepSeek V4 Flash.
Apple launched the next-generation Apple Intelligence today, with a Korean beta version of Siri AI set to roll out sequentially starting in October.
Samsung Electronics has released AnySimLite, an ultra-lightweight voice classification encoder capable of sub-30ms inference on smartphones with a size of 700KB.
Europe-based Desert Ant Labs has released 18 specialized small models, including audio and vision, that run on-device without cloud dependency.
Halo NeuroAI has released Sopro V2 Turbo, a 120M-parameter open-source voice cloning model that natively supports European Portuguese and enables real-time streaming on laptop CPUs.
Apple has unveiled the new Mac Studio featuring M5 Max and M5 Ultra chips.
Apple has unveiled the new Mac Studio featuring the M5 Max and M5 Ultra chips.
Apple has unveiled a new Mac mini equipped with M6 and M5 Pro chips, significantly enhancing AI and graphics performance.
By leveraging a Mixture-of-Experts (MoE) architecture, a 397B-parameter AI model has been successfully run locally on an iPhone.
Meta has released the Muse Glimmer model and ExecuTorch support for on-device agentic workflows.
This article introduces architectures and case studies for implementing on-device AI intelligence via external modules without modifying the firmware of existing robots.
Liquid AI released a 3.1B vision-language model with enhanced screen understanding and tool use.
Cactus has released Needle 2, a 14MB on-device agentic LLM.
NVIDIA supports on-device execution by quantizing ASR, TTS, and speech codecs to GGUF.
Liquid AI released a 2.6B agent model capable of running on mobile phones and CPUs.
Apple has unveiled a new audio synthesis architecture that generates high-quality speech in real time while minimizing memory usage.