glyd/Qwen3.5-9B-penguin

Qwen3.5-9B · Glyd penguin

12.7 GB, 29% smaller than bf16 (17.9 GB). Glyd's lossless compression of Qwen/Qwen3.5-9B (commit c2022362): every weight decodes to its bf16 value, bit for bit.

curl -LsSf https://getglyd.com/install.sh | sh
glyd run Qwen/Qwen3.5-9B:penguin

import glyd
model = glyd.from_pretrained("glyd/Qwen3.5-9B-penguin")   # pip install "glyd[gpu]"

Compared

	Size	Smaller than bf16	KL vs bf16	RTX 4090
bf16 (original)	17.9 GB	–	0	–
penguin (lossless, this repo)	12.7 GB	29%	0	–
kestrel	7.3 GB	59%	0.0147	122 tok/s
swift	6.2 GB	65%	0.0373	138 tok/s

Size: the files you download. KL: how far the model's next-token probabilities move from bf16, measured on WikiText-2.

Good to know

• Linux with an NVIDIA GPU (Ampere or later), driver 580 or newer; glyd 0.29.4 or later.
• Text only: the model's vision part is not included.
• The Hub's "8-bit precision" and "Model size" labels count packed bytes: this checkpoint stores about 11.4 bits per weight, from a model with 8,953,803,264 parameters.

License

Weights: Apache-2.0, from Qwen/Qwen3.5-9B. Glyd's engine, which runs them: BUSL-1.1, free for personal and non-commercial use on your own computers; commercial use needs a license. getglyd.com