KoboldCpp
Also known as: kobold cpp
KoboldCpp is a single-binary build of the llama.cpp stack, often used for long-form creative writing and roleplay.
- One file to run, no install.
- Ships a web UI and an API.
- Popular in the local “AI companion” community.
Niche but real: if you see "Kobold" in a local-AI context, it's this single-binary runtime plus its community of long-context writing tools.