#edge AI

'Nano Banana 2, reportedly running GemPix 2, could produce native 2K and upscaled 4K AI images on-device in seconds, signaling a shift toward studio-level mobile creativity with privacy and ethical trade-offs.'

'IBM launches Granite 4.0 Nano: eight compact models (350M and ~1B) in hybrid SSM and transformer variants, built for local and edge inference with enterprise-grade governance and open licensing.'

'Liquid AI launched LFM2-VL-3B, a 3B parameter vision language model built for edge deployments with efficient token compression, native aspect ratio support, and documented inference controls for predictable latency'

'NVIDIA introduced Jetson Thor, a high‑performance module and developer kit that brings server‑grade multimodal inference and generative reasoning to robotics at the edge. The platform pairs a Blackwell GPU with a focused software stack to enable real‑time perception, planning and action.'

'Alibaba's Ovis2.5 (9B and 2B) advances multimodal AI with a native-resolution vision transformer and an optional thinking mode, achieving top scores for open-source models under 40B and improved OCR and chart understanding.'

'A technical deep dive into AI inference in 2025, detailing latency bottlenecks, optimization methods like quantization and pruning, and a roundup of the top nine inference providers.'

Gemma 3 270M is a compact, 270M-parameter model by Google AI designed for energy-efficient, task-specific fine-tuning and on-device deployment with INT4 QAT support

'LEANN is a storage-efficient ANN index that recomputes embeddings on demand to cut index size below 5% while keeping strong accuracy and fast query times.'

Liquid AI announces LFM2, an advanced edge AI model series delivering faster inference and training, with a novel hybrid architecture optimized for deployment on resource-constrained devices.

Google DeepMind released Gemini Robotics On-Device, a local AI model that enables robots to perform complex tasks in real time without cloud connectivity, opening new possibilities for edge robotics.

OpenBMB launches MiniCPM4, ultra-efficient large language models optimized for edge devices, featuring sparse attention and rapid inference that outperforms competitors with less training data.

Iterate.ai raised $6.4 million to advance secure, scalable AI technologies designed to operate on enterprise edge devices, enabling privacy-first and cost-efficient AI adoption.

LuminX has secured $5.5 million in seed funding to advance its edge-based Vision Language Models, aiming to transform warehouse automation and improve real-time inventory insights.

NVIDIA introduces Llama Nemotron Nano 4B, a compact open-source AI model optimized for edge deployment that outperforms larger models in scientific reasoning and programming tasks.

PiLogic has raised $4 million to advance its exact AI technology tailored for aerospace and defense, providing precise, reliable AI solutions for critical edge applications.

Yubei Chen, co-founder of Aizip, shares insights on developing ultra-efficient AI models for edge devices and the future of autonomous edge AI powered by innovative automation.

PiezoMEMS technology is set to transform edge AI devices by overcoming key hardware challenges in thermal management, audio quality, and design, powering smarter smartphones and AR glasses.