Researchers propose MuCoDi, a pretraining framework that distills knowledge from multiple large pathology foundation models into compact edge-oriented encoders, achieving near-teacher performance with…
A practitioner is running Google's Gemma 4 E2B model on a single 4 GB VRAM card to handle screen watching, voice-memo and meeting transcription, and chat simultaneously, consolidating three previously…
A guide details how to build a private, offline AI assistant on a Raspberry Pi 5 using pre-trained language models, emphasizing the need for 8GB or 16GB RAM, active cooling, a proper power supply, and…
The Raspberry Pi 5 now comes in a 16 GB version for $120, enabling local AI model inference with MoE models like Qwen3.5 at 7–8 tokens per second, though real-world costs reach $180–220 with necessary…
Liquid AI released LFM2.5-230M, a 230-million-parameter model optimized for edge devices including phones, Raspberry Pi, and robots. The model achieves 213 tokens per second on a Galaxy S25 Ultra and …
Liquid AI released LFM2.5-230M, its smallest open-weight model optimized for on-device agentic tasks like data extraction and tool use, achieving 213 tok/s on a Galaxy S25 Ultra and 42 tok/s on a Rasp…
Liquid AI released LFM 2.5-230M, a 230-million-parameter model that outperforms larger models on data extraction benchmarks, achieving 22.51 on CaseReportBench versus 13.83 for Qwen3.5-0.8B and 2.28 f…
Squeez Labs demonstrated CrankGPT, a fully offline, hand-cranked AI voice assistant built on a Raspberry Pi 5 with 8GB RAM and a 20W generator, running local LLMs via llama.cpp. The device boots in 30…
SqueezLabs, a two-person company, has developed CrankGPT, a hand-powered chatbot that runs on a Raspberry Pi 5 with a hand-crank generator, designed for use in a post-apocalyptic scenario without grid…
A benchmark published on Hackster.io compares Seeed Studio's reCamera and Raspberry Pi 5 for edge vision AI deployment, testing inference throughput and setup complexity. The reCamera, built on Sophgo…
Researchers introduced RAMS, a runtime controller that dynamically switches between YOLOv8 model tiers on embedded devices to balance latency and detection quality under resource pressure. On Jetson O…
Katrin Tomanek and Alex Kauffmann built CrankGPT, a hand-cranked, fully offline AI box that serves as both a product satire and a serious edge computing benchmark. The device uses a Raspberry Pi 5, a …
A new smart device called AntiMould Shower Sentinel uses machine learning on an Arduino Q to detect shower sounds and check if an extractor fan is running, sending real-time alerts via Home Assistant …
Researchers have developed YOLO-AMC, an improved YOLO architecture integrating attention mechanisms for building crack detection, achieving mAP@0.5 of 0.9917 on test datasets. The model outperforms ba…
Squeez Labs released CrankGPT, a fully offline voice assistant powered by a hand crank and running on a Raspberry Pi 5 with 8 GB of RAM, according to the project website. The prototype uses a custom c…
IceWhale Technology's ZimaBoard 2 single-board computer, powered by an Intel N150 processor, is configured as a low-power Edge AI and IoT development server running Ubuntu Server 26.04 with Docker, Ju…
A developer's DoSync protocol solves the fragility of AI-smart home integrations by having each device publish a "Capability Manifest" that declares its own relevance to intents like emergency respons…
Developer Mohamed (MrKali) released ARN, a local semantic memory server for AI agents that runs on a Raspberry Pi 5 with 22ms recall speed and scored 10/10 in tests. The open-source tool stores agent …
Raspberry Pi 6 will not launch before 2028 and will not include an onboard neural processing unit (NPU) for AI tasks, according to three company engineers and leaders during a Reddit AMA on May 21, 20…
The article announces that Google DeepMind's Gemma 4 E2B model, released on April 2, 2026, can run on a Raspberry Pi 5 with 8 GB of RAM using only 1.5 GB of memory at INT4 quantization. This capabilit…