# @jmbollenbacher — 2026-09-16

♥1 ↻0 · https://x.com/jmbollenbacher/status/2100060215022330167

@ivanfioravanti have you done any experiments to push the quantization smaller? especially on the ngram embeddings?

it'd be cool to fit it into a 256GB setup. maybe weights ~200-210GB, like GLM-5.3?

not sure if its feasible with the embeddings table. but it'd be cool to see.

tags: author:jmbollenbacher, kind:tweet, on:glm-5-3, year:2026
cited on: _dossiers/glm-5-3.md, glm-5-3