~/runtime

Self-host an LLM

Five runtimes for serving open models on your own hardware. Pick by platform, throughput, or the kind of agent work you do. Each card opens a full install + features walkthrough.

Total
5runtimes
Cross-platform
2choices
Apple Silicon
2native
Server-grade
1production

Cross-platform

macOS · Linux · Windows · 2 options

Apple Silicon

macOS · M-series · 2 options

Server

CUDA · ROCm · production · 1 option