23 October 2026

Meta releases open 30B model that runs on one GPU

First reported

Deep Learning Weekly ran this on 23 October.

Meta has released Muse Glimmer 30B, a 30-billion-parameter AI model, under a permissive Apache 2.0 license.
The model is compressed to about 4-bit precision, so it can run on a single graphics card (GPU) rather than a data-center setup.
It was trained on text in more than 100 languages, and it is designed to carry out multi-step tasks on its own.

Our read

Only Deep Learning Weekly covered this release, first reporting it on 2026-10-23. It follows the 2026-10-20 story on Meta releasing Muse models while keeping its strongest one subscription-only.

Earlier on this story

Oct 20Meta releases Muse models, keeps its strongest one subscription-only
Oct 20Guide shows splitting one language model across two GPU servers on AWS
Aug 23Meta quietly became one of Microsoft's largest AI customers

Written by Newskeryx from which newsletters covered this story and from our own archive.

How it was covered

Deep Learning WeeklyEditorial team

Release license, quantization for single-GPU use, and multilingual training coverage