Updates

1. Infrastructure solved

   Research - even small models are either very RAM or GPU VRAM intensive, with 16GB of VRAM seeming to be the entry point Running locally on my gaming desktop 2. LLM provider selections made

   LM Studio hosting the gpt-oss-20b model locally

2. Core structure being built Choosing to leverage …

The What

I am participating in the OpenAI Open Model Hackathon (repo hosted here) hosted by DevPost for fun and some hands-on experience with self-hosting an LLM, in this case OpenAI's new gpt-oss model.

My project/ submission/ design is built around the idea of leveraging a local, self-hosted LLM to …