5 Quick Links for Devs: Week 7, 2026

February 13, 2026 · Jacob E. Dawson

New Model Extravaganza

Hot on the heels of Claude Opus 4.6 and GPT-5.3-codex come a couple of news models from Google and OpenAI. First is Gemini 3 Deep Think, which is available to Ultra subscribers and specializes in science, research and engineering. Then we've got the new Open AI's Spark, which is a smaller version of GPT-5.3-codex built for speed. I love 5.3 and will be trying out Deep Think this weekend!

Related: GPT-5.3-Codex-Spark

How I Built a Discord-like Chat App

Hampus Kraft goes deep into the creation of an open-source Discord-like chat app, including VoIP functionality and a really slick interface. It's neat to see the nuts and bolts of the app while also hearing about Hampus's challenges including marketing & promotion.

Using an Engineering Notebook

I used to maintain a TIL file in my github, and I had a lot of coding notes in Notion, but reading this post reminded me that I hadn't actually taken down any comprehensive coding notes for around 1 - 2 years. I'm not sure if it's because I default to discussing topics with coding models now, but it's prompted me to download Obsidian and try to get back into the habit!

Related: Obsidian

Setting Up Zed

I tried Zed a couple of times over the past year, but it never quite made it over the hump and I'm still using VSCode and occasionally AntiGravity. This little guide shares some initial setup tips to help get started, which I'll keep in mind when I try Zed again sometime later this year!

Related: Zed

Nobody Knows How the Whole System Works

In the midst of a future-panic around the coming LLM takeover I've been trying to dive into books on systems thinking, which is one of the core modes of thought that I think will be useful in the years to come as our role as developers changes. One of the core themes of systems thinking is that there is immense complexity within systems, and this blog post explores the example of modern technology, where things are so complex and interdependent that no single person actually knows how they work from top to bottom.