LLM Pigeon

So as the titles says I use this on a m3 Mac mini with 8gb of ram. It does exactly what it says it will do. Honestly this thing is great I use it with tailscale and LM Studio. I use tailscale so that I can use it on mobile data. I found it on reddit thanks to u/Valuable-Run2129. Honestly I am surprised they don’t make you at least buy the apps but hey I will take it. Oh NO SUBSCRIPTIONS!!!!

Nifty (free) tool and works! But really feels like a beta. Used it for about 5 minutes but I’ll try it out more. -I don’t want to download models in yet another place. I have things configured really well in LM Studio -You have to manually type a model name to interact with LM studio models, which is a shame because I’m certain LM Studio supports GET requests for model names. -You can only have one model from LM Studio configured at a time. Again, a GET request to LM Studio would fox this. -I would love to see a chat stream, but I understand if that’s a limitation of architecture requiring minimal setup. Fair trade off if so. -I would like the option to see thoughts from thinking models -If a model isn’t loaded in LM Studio, it gets stuck in a loop if a chat is requested. I think there is a setting to load models on demand in LM Studio, but some feedback from Pigeon would be helpful for troubleshooting.

I love the way this app works with LM Studio on a Mac to enable us to access our local LLMs via iCloud. Companion server for MacOS needed so be sure to download that. Open source and free! What more could we want?