Model reference · open weights

xglm

LLMs facebook Text gen 1 build Open weights 102k dl/mo

xglm is an open-weight language model from Meta. xglm-564M (BF16) weighs 1.1 GB; the smallest configuration that runs it is RTX 3060 12 GB.

XGLM-564M is a multilingual autoregressive language model developed by Meta with 564 million parameters.
It is designed for text generation tasks and supports a context length of 2048 tokens.
The model was trained on a balanced corpus of 30 languages totaling 500 billion sub-tokens and is released under the MIT licence.

Summary of the facebook/xglm-564M model card, 2026-10-01

What it is

What it runs on

Memory and cards for xglm-564M (BF16)

1.1 GBweights, file size

742 MBruntime overhead, at least

2,048 tokenscontext max

How much memory each request adds isn't estimated yet for this architecture. The weights need at least the cards below, plus room for the context.

Details

Languages

multilingualenruzhdeesfrjaitptelkofiid