Kun Chen is amazing! He did a Twitter post that I transcribe bellow: "just published a free OSS tool "runner-up" github.com/kunchenguid/ru… which is the setup i've been using to self-host all my github actions runners across my own machines i run thousands of CI jobs every day on private repos. they would have cost me about $5000-$9000/mo depending on what cloud provider i choose, which is completely bogus because that's a new mac studio every month after lots of calculations it was clear that simply owning my own hardware would be an order of magnitude more economical, which led to me creating this setup and i finally got some time to get it published if you have a similar need, hope this helps!" Check the repo, it's amazing and it solved a big problem for me. https://github.com/kunchenguid/runner-up

How cool is this? There goes my weekend lol https://youtu.be/mrdFgLXiy-o?is=0Ni_zjqC6-SxCAs9 https://github.com/wombatoperator/omarchy-voice

I keep seeing people complain about usage limits with OpenAI and Claude models. I used to hit those limits too. Am I an expert coder? No. I build a lot of projects, though, and I rely heavily on coding models. So I tried a Devin subscription. It’s been one of the best decisions I’ve made for my workflow. SWE-2’s usage feels close to unlimited for the kind of work I do, but the amount of usage is only part of it. What surprised me was how well it fits into a workflow with other models. I ask Astra or Fable to think through a task and write a detailed implementation plan for a slightly weaker model. I want the important steps spelled out, with as little left implicit as possible. Then I take that plan to Devin CLI and ask SWE-2 to implement it. I’m not a huge fan of the Devin app itself. The CLI is where it clicked for me. He’s such a good boy. GIVE IT A CHANCE! It follows the plan, keeps working for hours without losing the thread, and handles implementation problems the planning model didn’t anticipate. It doesn’t need to be the most brilliant model in the room. For this job, I need it to be technically good and consistent, and it is. At first, I sent its work back to Astra or Fable for a careful review. Had it followed the plan? Missed anything? Was there something they would change? Their assessments were positive again and again. Eventually, I stopped sending every result back for that second pass. That’s probably the biggest compliment I can give a coding model: I trust it to execute. This also means I’m not spending frontier-model tokens on hours of file edits. I use those models to reason and plan, then give SWE-2 the sustained implementation work. Even after the current promotions end, I think that combination of cost, reliability, and follow-through will be attractive. I currently pay for Claude, OpenAI, and SuperGrok. More and more, I’m considering replacing Grok with Devin. I’d miss Grok Bot most, and I hope OpenAI eventually gives us something comparable. But for what I do every day, Devin is giving me more value.

So... I told Astra: build me an high quality game, fast paced and fun, to teach me all about git. You are in charge, use blender and 3js, your goal is to complete the game. I'm going to sleep! Let's see!

It quite surprised me that it's IBM doing this :) For anyone lost in the sea of terminologies, principles and needs guidance (like me), I recommend the channel bellow. IBM tutors go through different subjects in a clear and short way. https://youtube.com/@ibmtechnology?si=Mfq1iP1VVHlTf8b2