Search fresh public links, source activity, and ready-to-use post angles for Tinyml.
Fresh curated links around TinyML are collected here so marketers can spot useful updates and turn timely ideas into posts faster.
Recent items include:
Recent curated links from global sources. Generate one free draft from any story, then use SocialBu to schedule and refine your content calendar.
Let’s be honest.Continue reading on Medium »
Liquid AI released LFM2.5-230M, its smallest model yet. The 230M-parameter, open-weight model runs on-device at 213 tok/s on a Galaxy S25 Ultra and 42 on a Raspberry Pi 5. Built on...
Liquid AI, founded by former MIT computer scientists, today released its smallest AI language model yet, LFM2.5-230M, and enterprises would do well to consider it for their uses in...
While Android already supports Local Gemini Nano via AICore, only a select few devices are currently included in the supported list.Continue reading on Medium »
What if a single edge AI platform could capture voice in noisy environments, detect machine faults, and bring AI to wearables, drones, and industrial systems? Upbeat Technology has...
Fine-tune open language models locally on your Mac using MLX. No cloud GPUs or costs required.
Fine-tune open language models locally on your Mac using MLX. No cloud GPUs or costs required.
Gemma 4 models now use quantization-aware training to use less memory while retaining quality performance.
Ambient Scientific GPX10 Pro is a processor/microcontroller for always-on embedded AI applications on power-constrained edge devices. With a consumption of under <100μW, it can...
Pete Warden is convinced local voice interfaces and sub-$1 embedded chips will fundamentally change how we interact with everything in the physical world. I’m so excited to introdu...
MiniMax M3 introduces MiniMax Sparse Attention, a 1M-token context window, and native image, video, and computer use support. The post MiniMax Releases MiniMax M3 with MSA Architec...
В прошлый раз я рассказал про GraphKAN, первую тернарную KAN с весами {-1, 0, +1}, которая выдала 96.15% на MNIST при 15 КБ. (ссылка).С тех пор проект переродился. Рассказываю, что...
MiniMax released its M3 flagship model, claiming it as the first domestic AI model to combine frontier coding, agentic capabilities, 1M-token context windows, and native multimodal...
On my Pixel, the most compressed model I built ran slower than the bigger one, and it thought a dog was a shower curtainContinue reading on Medium »
An edge AI sensor platform combines motion, sound, light, and environmental sensors with on-device AI to process data without cloud connectivity. The TIDA-010997 Edge AI Sensor Boo...
ZML, a hot French AI startup endorsed by Turing Award winner Yann LeCun, has now released ZML/LLMD, software that could make running AI less costly.
A system-on-module brings AI processing, video analytics, and machine learning to edge devices, helping engineers build robotics, surveillance, and automation systems. Calixto Syst...
If you read my previous article on on-device AI in Android, you already know why running models locally matters: faster inference, better…Continue reading on Medium »
Chinese AI unicorn MiniMax is preparing to launch its M3 large language model featuring a custom sparse attention mechanism, claiming 9.7x prefilling speed improvements.
How I deployed a state-of-the-art object detector on the Kendryte K230 for industrial textile monitoring using INT8 quantization, nncase…Continue reading on Medium »
MacKenzie Sigalos / CNBC: PrismML launches Bonsai 27B, a model based on Qwen3.6 27B that it says runs natively on Apple devices via MLX; its CEO says Apple is evaluating the tech ...
Liquid AI's LFM2.5-8B-A1B activates 1.5B of 8.3B parameters, offering 128K context, reasoning, and tool calling on consumer hardware. The post Liquid AI Releases LFM2.5-8B-A1B: An...
New AI compression technology targets always-on devices by significantly reducing memory and transmission overhead, helping wearables and edge systems extend battery life while ret...
Xiaomi's MiMo team, with TileRT, released MiMo-V2.5-Pro-UltraSpeed, a serving mode for the MiMo-V2.5-Pro model. It decodes over 1000 tokens per second on a 1-trillion-parameter mod...
Use SocialBu to discover ideas, generate post drafts, and schedule them across your social channels.