Want to run Hermes locally but have no idea which model your computer can handle? Already running local models but want an easier way to choose, download, and manage them? Hermes Desktop can now help with both. Local models are becoming a first-class provider inside Hermes. Hermes maintains a curated catalog of supported local models and can compare those options against the hardware actually sitting in your machine. Launch Desktop with: `hermes desktop --local` Then go to: Settings → Providers → Local models Hermes checks things like your: • GPU • VRAM • RAM • unified memory Then it can rank the available models by what your machine can realistically handle, recommend a model and the right build for your hardware, and explain why it picked it. If you are starting from zero, there is a quickstart that can handle the rest: • install the llama.cpp runtime • download the recommended model • start the local server • make the model available inside Hermes No hunting for GGUF files, guessing which quantization you need, or manually wiring everything together. And if you already have models downloaded through Hermes' local setup, Desktop can also show what is installed, what is loaded into memory, what is active, and let you manage them from the same place. Downloaded local models can then appear alongside your other providers in the normal Hermes model picker. So the new flow is much closer to: Your hardware → Hermes recommends what fits → download → run locally. And for people already running local models: Your models → Hermes helps manage them.