Run LLMs locally with one command
One command downloads and serves models with an OpenAI-compatible API — privacy first.
Cloud inference and fine-tuning for open models
One API, hundreds of models
A desktop GUI for local models
Per-function cloud GPUs with seconds of cold start