Gotcha, I think you can configure models and params with ollama, think that's possible here, or would need to go the more advanced llama method?
02.03.2026 15:20
👍 0
🔁 0
💬 1
📌 0
Gotcha, I think you can configure models and params with ollama, think that's possible here, or would need to go the more advanced llama method?
Is that possible with ollama? I'm still new to local models and ollama makes that easy for me still. I saw there is Qwen3.5 ollama.com/library/qwen... and tried the lowest one on a M4 but was still slow out of the box.
Jumped from 20 TPS to 80 TPS ⚡ on my M4 Pro.
The trick is that this version uses a MoE architecture with 3 billion active parameters. That makes it way faster.
I was running the base model via GGUF (ollama), and now running on MLX (vlm - because this is a vision model too).
Just deleted five 1-1s from my calendar for this week since none of them work with me anymore. 😬
Wow! Where are you able to host this? Or are you using Firebase Test Lab for the device? That's amazing.
I think I'm going to be blogging more about it at some point here. Will be sure to share when I do but basically every plan had a number prefix like 0019-...MD and had metadata as frontmatter
Gemini just gave me feedback: "This is arguably the most rigorous, agent-friendly documentation system I've seen. Your repository is now fully structured for rapid, autonomous development." - We are all totally having to change the way "we" work to make "them" effective. 🤖
Really nice episode on the history of Kotlin and the future of programming languages by Andrey Breslav & @pragmaticengineer.com https://pca.st/episode/8d6081d6-3419-4e9a-bbf7-332257cf7da2
handstandsam.com/2026/01/07/h...
Will need to do a follow up for more recent. Mainly using sub agents and Opus 4.6
Same, but it's been happening for all parts of the day. Before making dinner or going to bed, etc. 😳
So true. I just posted bsky.app/profile/hand... and immediately saw your post after. Hope you enjoy the long weekend.
Yes, this!
The last month has been a non-stop (nights & weekends) wild ride of coding with agents and shipping. I've been genuinely excited and learned so much. Definitely am burned out, and glad for the long holiday weekend with fun things to do and no computer. So many ideas that will have to wait. 😂
Matterport probably. Been the standard for real estate a while and has a phone app to do it.
Valentine 💘 or openclaw.ai 🦞
ChatGPT spotted in the Barnes and Noble magazine section.
Estimated timelines showing how long this will take in person-days
So... this work got done in under an hour using this strategy w/Firebender and multiple parallel agents using Claude Opus 4.5. How long will it be before these estimates get updated to new agent development strategies in these models? 😱🤖
I'm at about 12 "You're absolutely right!" messages today from Claude. I guess I've still got a job for a little while.
@mikepenz.dev Thank you for your Compose multiplatform material Markdown library. I'm using it in a couple places. It's great!
Learning about Open Telemetry Specs. Always heard of it, but never used it. This viewer was helpful to visualize: github.com/CtrlSpice/ot...
Ktor 3.4.0 is out! 🎉
The release improves stability and brings new features:
🔶 Zstd compression
🔶 OkHttp duplex streaming
🔶 Structured concurrency for the HTTP request lifecycle
🔶 OpenAPI generation from code
Read the blog post for details ⬇️
https://kotl.in/n31gtm
Ktor is the best!
Oh yeah, wow. Server stuff is a whole different knowledge set than client-side work.
I always thought it was CloudFare from hearing people mention it, but it's CloudFlare it seems. Just in case anyone else didn't realize that either... 🫣
@johnoreilly.dev You were mentioned as Firebender was trying to upgrade to the latest Compose Multiplatform and Navigation 3. 😲
Much better answer than mine 😂
I understand the feeling. Lately I'm letting it make me faster and I steer and correct it.
