I specifically chose PrismML's 1-bit models because their tiny size allows them to actually fit on smaller hardware like the Pi. The 1.7B model is great for basic tasks and tool triggers, while the 4B model seems reasonable for some daily tasks, though it's much slower on this setup. If you try these models on your M1 Max, I assume they'll run incredibly fast. I previously tried them on a VPS and the inference speed was really good for my experiment.
I spent the weekend experimenting to see if I could get a proper LLM running locally on an old Raspberry Pi 4 (4GB), and more importantly, if I could get it to interact with the physical world.
I ended up using PrismML's new Bonsai models. Because they are genuinely 1-bit (trained from scratch at 1-bit, not quantized down to 4-bit), they actually fit. The 4B parameter model is ~570 MB, and the 1.7B is ~240 MB.
I loaded them through llama.cpp's router mode. I get around 2 tok/s on the 4B model for better reasoning, and 4-5 tok/s on the 1.7B when I just need speed. I tried Gemma 4 E2B first, but it was just too slow on 4GB of RAM.
The fun part: I wired up a cheap TM1637 4-digit display to the GPIO pins. Since Bonsai supports native tool calling, I wrote a small Python proxy that injects an update_display function into requests. When the model decides to use the tool, the proxy catches the streaming call, extracts the text, and drives the display. You can tell it to "show 1453" and it physically lights up.
It’s definitely just a weekend project (7-segment displays can't render W or M, self-signed certs, etc.). The code and setup scripts are all in the repo.
I’m thinking about adding servos or sensors next. Would love to hear your thoughts or see if anyone else is building edge AI hardware projects!
Is there a place where you have added the instructions on how to set it up? I was trying it out but getting server resources issue for 8B on my 4GB machine.
I've been experimenting with AI Agents to see how far I can push them across different types of complexity (from dashboards to client-side video encoding).
Trip Replay (https://tripreplay.app) - A client-side travel map animator where I successfully got the AI to implement complex D3 projections and WebCodecs logic.
Krypto Markets (https://krypto.markets) – A crypto dashboard built purely in "Agent Mode" to test how fast I could ship a data-heavy UI.
Gez.la (https://gez.la) – My old COVID-era open source virtual tour database project that I used agents to fully refactor and modernize from a legacy stack.
I just launched Trip Replay, my last project of 2025.
There are already tools that create animated travel maps, but they are almost all paid apps. Some charge per export, others require a subscription. I wanted something I could use myself without friction. So I built a free, web-based alternative.
How it works: It runs entirely in the browser using Next.js and the Canvas API.
* Rendering: It draws the map (using D3-geo) and the path frame-by-frame on a hidden canvas.
* Encoding: Instead of a server-side render, it uses WebCodecs to encode the video directly on the client.
* Privacy: Since it's client-side, no data leaves your device (except location search via API).
The output is a 1080p vertical video (9:16) ready for social media.
I wanted to test how far I could push AI-driven development. Claude handled WebSocket integration, responsive design, chart implementation, auth, and debugging. My role was mostly directing, reviewing, and making design decisions.
The result surprised me. What I expected to take weeks was functional in under 48 hours. Not everything was perfect on the first try, but the iteration speed was remarkable.
Curious to hear HN's thoughts on AI-driven development and the project itself.
reply