Arrakis / Models / diffusiongemma-26B-A4B-it

diffusiongemma-26B-A4B-it GPU requirements

diffusiongemma-26B-A4B-it has 26B parameters. To serve it, plan for about 62 GB of GPU memory in 16-bit, 31 GB in 8-bit or 15 GB in 4-bit, including about 20% for the KV cache and activations.

It is a mixture of experts: 128 experts, 8 of them active for each token. All of them stay in GPU memory, so plan memory for all 26B parameters.

Deploy this model Check another model

What it takes to serve it

GPU memory to run it for inference, with the smallest GPU setup that holds it and the lowest price on Arrakis right now.

Precision	Memory and the smallest setup
16-bit	62 GB: one H100 or A100, H100 from $1.65 per GPU-hour
8-bit	31 GB: one RTX 5090, from $0.58 per GPU-hour
4-bit	15 GB: one RTX 4090 or RTX 3090, RTX 3090 from $0.19 per GPU-hour

What it takes to fine-tune it

QLoRA trains adapters on a 4-bit frozen model, LoRA on a 16-bit one, and full fine-tuning updates every weight. Figures are before activations, which grow with batch size and sequence length.

Method	Memory and the smallest setup
QLoRA	18 GB: one RTX 4090 or RTX 3090, RTX 3090 from $0.19 per GPU-hour
LoRA	62 GB: one H100 or A100, H100 from $1.65 per GPU-hour
Full fine-tuning	413 GB: 2 × B300, from $9.20 per hour for all 2

About the model

From its card and files on Hugging Face. Memory figures here follow the GPU memory guide.

Parameters: 26B
Active for each token: 4B
Experts: 128
Experts for each token: 8
Context window (tokens): 262,144
Published in: BF16
License: apache-2.0
Released: 2026-06-09
Weights: google/diffusiongemma-26B-A4B-it

Deploy it on Arrakis

Open Models in the console, choose vLLM, SGLang or llama.cpp, and rent a GPU it fits on. You get an OpenAI-compatible endpoint with its own API key, and root SSH to the machine. Or run it yourself on any machine with enough GPU memory:

vLLM

vllm serve google/diffusiongemma-26B-A4B-it --tensor-parallel-size 1

Deploy this model How deploying works

Questions

Live prices for every GPU are on the GPU prices page.

How much GPU memory does diffusiongemma-26B-A4B-it need?

To serve it, about 62 GB in 16-bit, 31 GB in 8-bit, 15 GB in 4-bit, including about 20% for the KV cache and activations. To fine-tune it, about 18 GB with QLoRA, 62 GB with LoRA and 413 GB for full fine-tuning with Adam, before activations.

Can I run diffusiongemma-26B-A4B-it on one GPU?

Yes. In 16-bit it needs about 62 GB, which fits on one H100 or A100.

What does it cost to run diffusiongemma-26B-A4B-it?

In 16-bit it needs one H100 or A100; in 4-bit, one RTX 4090 or RTX 3090. Each setup on this page shows the lowest price per GPU-hour on Arrakis right now, and you pay by the hour from a prepaid balance.

How do I deploy diffusiongemma-26B-A4B-it on Arrakis?

Open Models in the Arrakis console, paste google/diffusiongemma-26B-A4B-it, choose vLLM, SGLang or llama.cpp, and rent a GPU it fits on. You get an OpenAI-compatible endpoint with its own API key, and root SSH to the machine.

More models

Every model, and a check for any other.