* [llama-swap v198](https://github.com/mostlygeek/llama-swap) – Reliable on-demand model switching between local OpenAI-compatible inference servers without restarting applications.