Model reference · open weights

medgemma-text

LLMs google Text gen 1 build Its own licence terms 25k dl/mo

medgemma-text is an open-weight language model from Google. medgemma-27b-text-it (BF16) weighs 54.0 GB; the smallest configuration that runs it is H100 80 GB.

MedGemma 27B is a text-only large language model developed by Google, built on the Gemma 3 architecture and trained exclusively on medical text.
It is designed for text generation tasks in healthcare applications and features a 27.0B parameter size with a context length of at least 128K tokens.
The model is available as an instruction-tuned version and is governed by the Health AI Developer Foundations terms of use.

Summary of the google/medgemma-27b-text-it model card, 2026-10-01

Hugging Face ↗GitHub ↗Paper ↗

What it is

Released by	Google
Released	2025-05-19
Parameters	27.0B
VRAM	54.0 GB for the weights

What it runs on

Memory and cards for medgemma-27b-text-it (BF16)

54.0 GBweights, file size

762 MBruntime overhead, at least

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.

Card	Weights alone
RTX 3060 12 GB … L40S 48 GB 6 smaller cards	does not fit
A100 80 GB	fits
H100 80 GB	fits
RTX PRO 6000 Blackwell 96 GB	fits
DGX Spark (GB10) 128 GB unified	fits
H200 141 GB	fits
B200 180 GB	fits

How it works

How language models work

Details

Type	Language models
Task	Text gen
Runs with	transformers
Based on	google/gemma-3-27b-pt
Popularity	25k downloads / month
Weights	54.0 GB, medgemma-27b-text-it (BF16, file size)