A 3B-active open model is suddenly useful locally.
Opus 4.7 may cost more per session.
A big breach you should know about.
Codex wants to control your desktop.
A cloud migration that saved about $14k/year.
All that and more in this week’s AI newsletter:
www.linkedin.com/posts/jeremy...
This is an awesome tool. You can even run it local on your machine in just a few steps. Drop in a GitHub URL and it will visualize the application for you.
Check it out here --> github.com/braedonsaund...
This is a visualization of my ham radio test practice app.
I don't know when "Tell me a funny joke about Python" became my first test of a new model, but it's been working for years. Trying out Qwen 3.6. Subsequent tests will be more thorough
Karpathy sketched the gap. Graphify turned it into a CLI. A persistent knowledge-graph layer for coding agents that claims up to 71.5x fewer query tokens on mixed corpora.
github.com/safishamsi/g...
A new paper removes the CPU from LLM inference entirely. Blink uses a SmartNIC to deliver inputs into GPU memory via RDMA while a persistent GPU kernel handles batching and scheduling.
arxiv.org/abs/2604.07609
Coders in 2026
A cuBLAS bug is causing RTX 5090 GPUs to use only about 40% of available compute on batched FP32 matrix multiply workloads. If you bought one for local AI work, you are getting less than half the performance you paid for. NVIDIA has not acknowledged it publicly yet.
www.cloudrift.ai/blog/beating...
GitHub Copilot CLI now lets you run two different foundation models on the same task. Sonnet generates the plan, GPT-5.4 reviews it before execution. The combo closed 75% of the performance gap on complex multi-file changes.
github.blog/ai-and-ml/gi...
Anthropic chose not to release its latest model after it found thousands of vulnerabilities in popular software, some dating back 27 years. The US Treasury gathered banking leaders in Washington to discuss the cybersecurity fallout.
www.theguardian.com/technology/2...
Anthropic shelved a model after it found thousands of legacy zero-days. The US Treasury called bank execs to Washington. Also this week: GitHub Copilot CLI now cross-checks with a second model before executing. New issue is live.
www.linkedin.com/pulse/anthro...
The Schema Is the Product: An Architectural Reading of Karpathy’s LLM Wiki
medium.com/@han.heloir/...
This is the best lightning video from the Portland area storm this weekend. My wife filmed this from the couch! This was going on in our front yard.
www.facebook.com/share/r/18TZ...
I am blessed. I have a fast machine with an RTX4090 and a bunch of RAM. During the day it runs Linux and I use it to work mostly thru SSH. I've build so much with it.
I reboot it into Windows and it becomes a incredible gaming machine for competitive sim racing!
Poor thing never gets a break.
Super happy with my latest Linux Foundation @linuxfoundation.org certification. I would definitely recommend it, it covers some important topics. AMA
9 RAG Architectures Every AI Developer Must Know: A Complete Guide with Examples
pub.towardsai.net/rag-architec...
If you're building stuff with Langchain, you need Langsmith tools. Even if you use it for nothing else, monitoring costs in near realtime is worth the few minutes to set it up.
this is me for sure
Engineering teams in SF are moving from conversational AI coding to "harness engineering": bounded, observable loops where agents pick up test writing and refactoring overnight, presenting PR candidates by morning.
escape.tech/blog/everyth...
An open-source library maps agent actions directly to EU AI Act Articles 9 through 15 using HMAC-SHA256 signed audit chains. The August 2026 enforcement deadline is coming and your existing JSON logs will not satisfy regulators.
github.com/airblackbox/...
