BotPsy: On-Device Personal Assistant with a Sub-Billion-Parameter Agentic Language Model
huggingface.co — A Blog post by QVAC on Hugging Face
Small Models
Small models are artificial intelligence models designed for low parameter counts and efficient resource footprints to run locally on devices. Truncated covers their efficiency techniques, hardware compatibility, and deployment strategies.
A Forgotten 1988 NASA Design, And The Model That Reads Ten Million Tokens In 25 Megabytes
brianroemmele.substack.com — The Memory That Never Grows
KORA builds AI models and agents into a new Linux OS bound for Qualcomm Snapdragon PCs
siliconangle.com — KORA builds AI models and agents into a new Linux OS bound for Qualcomm Snapdragon PCs - SiliconANGLE
Presis Brings Self-Hosted Classification to Boost.ai’s Agent Platform
www.unite.ai — boost.ai announced the launch of Presis on November 10, 2026, a self-hosted language model for classification decisions around AI agent conversations. The company said internal testing found it at least 2.5 times faster...
Meta’s Muse Ember wins on cost of agentic tasks
thedeepview.com — Meta has a new AI model that CFOs are going to love—and not because it automates spreadsheets. On Monday, Meta launched Muse Ember, its budget-tier AI model to...
Muse Ember (Max) from @AIatMeta just landed in the Code Arena: WebDev as the highest-performing small model - built for…
Meta launches Muse Ember to take on Claude Haiku and GPT-6 Luna
thenewstack.io — Meta’s Muse Ember is a low-cost multimodal AI model for coding and computer use, with benchmark scores close to Claude Haiku 5.5 and Meta’s Muse Spark 1.3.
I stopped chasing bigger local models, and a 27B model on my 8GB GPU does more of what I need
www.xda-developers.com — I don't download every new model anymore
When To Build Your Own AI Model (And When To Stop Kidding Yourself)
www.forbes.com — Post-training and SLMs are real, powerful tools, but they work best as engineering and business decisions, not marketing stories.
Pinterest Replaces Journey Clustering with 4B Qwen3 LLM, Reports Improved Engagement
www.infoq.com — Pinterest has moved its user-journey inference system from a multi-stage clustering pipeline to a fine-tuned 4-billion-parameter language model, reporting a 1.1% improvement in email click-through and 1.3% higher push opens.
The Score That Cannot Be Gamed.
brianroemmele.substack.com — In 1988 Pentti Kanerva published Sparse Distributed Memory from NASA Ames Research Center under contract NCC2-408.
A Forgotten 1950 Rain Formula, And The Open AI That Only Picks
brianroemmele.substack.com — FEBRUARY 1950
What Is Model Distillation? How Large AI Models Teach Smaller Ones
Unite.AI — Model distillation transfers behavior from a larger or more capable teacher into a smaller student model using teacher outputs, internal signals, or softened probability targets. This guide explains the mechanism, trade-...
Someone got an offline LLM running on a 2015 Kindle Paperwhite, even if it’s quite slow
XDA — Okay, it's very slow. But it's still impressive.
LightOn tells LDS why finding the right file is not enough
Let's Data Science — LightOn's Maxence Lasbordes tells LDS how a local search agent found the right xarray file but missed the fix, when delegating code search pays off and why local search alone does not keep code private.
Thinking Machines' new model takes the lead in western open-weight AI
ZDNET — Another competitive model developed outside of China could change the open-weight landscape.
Underdog shows off a 32MB speech model in new demo
TechCrunch — Despite its small size, Underdog's new model can generate both speech and music.
Exciting news: Inkling 2 Small by @thinkymachines has landed in Code Arena: WebDev with 1654 pts (AutoEval).
Dharma Field Extractor: Any Document In, Structured Fields in JSON Out
huggingface.co — A Blog post by Dharma-AI on Hugging Face
What happens when AI starts self-replicating?
Engadget — Anthropic warns that future is soon, but the threat is less novel than it seems.
Dismiss email signup
Subscribe to 
The top AI news in your email delivered each morning. Choose categories