What's next for Glyd: pip install, vLLM, one command

Everything this site shows was measured with Glyd’s PyTorch harness: clone the repository, set up the environment, and run e2e.py on a model you downloaded (getting started). That is the honest way to measure, and a slow way to use it. Here is the plan to make it one command, in three phases. They are plans in progress, with no dates.

1. pip install

pip install "glyd[gpu]"

A package for the GPU part, and models already packed on Hugging Face, so that loading one is a line of Python with glyd.from_pretrained(): the model in about 11 bits a weight on the GPU, every weight its bf16 value, without packing it yourself.

2. Serving through vLLM

vllm serve MODEL --quantization glyd

Serving engines are how models run in production, and today Glyd is not in one: its speed numbers come from the harness, with q, k, v and gate, up merged the way vLLM runs them. The second phase is Glyd as an option of vLLM, so that a model that needs two GPUs in bf16 can be served from one, as Qwen3-32B runs on one 48 GB GPU in the harness.

3. One command

glyd fit
glyd pack
glyd serve

The steps of the harness as commands of their own: whether a model fits your GPU, the question the GPU pages answer from the fit rule; packing it; and serving it.

What stays the same

The model does not change: every weight decodes to exactly the bf16 value it was, so the answers are the model’s, up to the order a GPU adds its products in (why). And every number keeps coming from a run whose log is in the repository: benchmarks/gpu.

Until then, the harness runs any model released in bf16 today. What changes as it happens: the changelog and the roadmap, and this blog’s RSS feed.