Artifacts, but 100% Open Source, Private & Local. Open WebUI has released a game changer update---"Artifacts". Instead of proprietary LLMs, you can now use fully LOCAL and PRIVATE LLM via ollama to build apps! Here I use Qwen 2.5 32B to build a game of life app.

Open WebUI supports ALL OS and platforms. You can easily install and run it locally with 1-click with

Here I took the original game of life app, and asked the AI to use smile emojis 🙂 instead. Watch what happens!

Then I asked it to use 40x40 grid instead of 100x100. Then 20x20 😊

Finally, instead of looking like pixels moving around, I wanted the whole thing to actually feel like looking at a life form through a microscope. At first I failed to fully convey my message, but eventually I got it to work in the end (Watch till the end!)

The artifacts feature ships with version control so you can sift through multiple versions you've created along the way.

Tips 1. Try to use the highest quality model you can fit in your VRAM (Qwen 2.5 32b is 20GB, which worked with NVIDIA 4090 with 24GB vram) 2. When building a web app, ask the AI to give a single HTML file. Otherwise it may give you multiple split files which may not work.

So this is kinda crazy. Love this new feature you showed us and so, I went and installed the new @Openwebui. I have a standard test I do for app dev in LLMs and it is to build an incremental timer for breaking bad habits like smoking. Anyway, I was interacting with it normally and asking for changes with each version and loving the ability to try out the code in the UI. Then suddenly, Qwen 2.5 32B added a new "assistant" character to the conversation and proceeded to go off the rails creating multiple versions of the app in one chat and adding functions I never needed. It went so far as to ask for code that would handle millions of digits in a number and then went into an elongated explanation of how the code would work and then added a mysterious human to the prompts that asked it to solve a work problem involving penguins and Kiwis. Watch the video, but this was kinda crazy. I've seen some bizarre behavior with LLMs, but I think this one wins the prize for adding characters to the conversation that redirected it with zero prompting by me.

@OpenWebUI @ollama @Alibaba_Qwen thank you for sharing! really cool

@OpenWebUI @ollama @Alibaba_Qwen Holy crap it works good. Prompt: Create me a single page website about guinea pigs.

@OpenWebUI @ollama @Alibaba_Qwen this is huge release because the most loved feature of artifacts can be now be utilized with any model running locally