← sources

Quartermaster 1.7 update now lets you run llama.cpp's new decision models (/v1/systemone) with a playground tab

source: reddit (reddit:LocalLLaMA)published 2026-10-13 17:54ingested 2026-10-14 02:43

retained text


Quartermaster (my local inference manager: one server that swaps llama.cpp, image, audio and video models in and out of VRAM) can now help you server and manage decision models:

- **Auto-discovered.** Drop a decision GGUF (Kev, OpenJev, Laya, ...) in your models folder and it gets its own config entry, sized for VRAM like everything else.
- **Playground tab.** Write a prompt, ask one question, pick the answer shape, hit Run. Each option gets a probability bar. Useful for quick testing
- **Model browser.** A Decision tab finds them on Hugging Face.
- **Same API, proxied.** `POST /v1/systemone` goes through Quartermaster's scheduler and API keys, so it swaps in next to your chat model like any other request.

Needs a llama.cpp new enough to have `/v1/systemone` (upstream b11361+; Quartermaster will warn about older builds).

Project info: https://quartermaster-labs.github.io/Quartermaster/
Release: https://github.com/Quartermaster-Labs/Quartermaster/releases/tag/v1.7.0

Happy to hear what people are using these for.

document fetched 2026-10-14 00:42 · canonical https://reddit.com/r/LocalLLaMA/comments/1x53ob3/quartermaster_17_update_now_lets_you_run