Run Qwen’s 27B vision model in Ollama at a size you pick

Made by IST Austria Distributed Algorithms and Systems Lab

Compressed copies of the Qwen3.8-27B model at four different file sizes, made by a university research lab, which run unchanged in llama.cpp, Ollama and LM Studio.

Each one includes the extra piece that lets the model read images.

You could use it to…

Pick the file size that actually fits your GPU
Run a vision model through Ollama or LM Studio
Read images with a compressed copy of Qwen3.8-27B

Can I use this?

Runs via Ollama or LM Studio

You'll need Ollama, LM Studio or llama.cpp · Setup needed

Worth knowing These are compressed copies of someone else’s model, so what it knows and how it behaves come from the original. Choosing between the four sizes is a trade-off you have to make, and the quality figures for each are the lab’s own. Released under Apache 2.0, following the original model. BuildTube has not run or verified this model.

Why it's here

Getting attention on Hugging Face

1,866 people have liked it on Hugging Face.
It was downloaded 1,679,425 times in the last 30 days.

Numbers from the snapshot taken 1 October 2026; not refreshed since.

Behind it

See the code on Hugging Face

BuildTube has not run or verified this project. Everything above is written from what the creator published.

Made this?