# Built a VRAM scheduler (vramble) to automate model swapping on my local AI workstation — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/9d720174c88c7aee

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-13 05:28 ingested 2026-10-14 02:43

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1x4ocgn/built_a_vram_scheduler_vramble_to_automate_model/)

## retained text



```
I set up a dedicated workstation for local AI experimentation with various models and tools. But every single one of them wants all the available VRAM.

To fix this, I built a request dispatching layer called **vramble**. It automatically loads and unloads models and manages the queue using a priority system. Now I can run different workflows without manually managing models (or wasting Claude tokens to orchestrate the process).

Any contribution is welcome, it works for me but is a new build so can't guarantee that will work directly on any machine

Side note: Qwen 27B at 200+ t/s is performing way better than I expected. opeMIT license, available on GitHub: [github.com/ValerioDolci/vramble](http://github.com/ValerioDolci/vramble)
```



document fetched 2026-10-14 00:42 · canonical https://reddit.com/r/LocalLLaMA/comments/1x4ocgn/built_a_vram_scheduler_vramble_to_automate_model
