Exhibit · AI art generation
Generate AI art with real GPU power
16 GB to 80 GB of VRAM. Run Stable Diffusion, fine-tune LoRAs or run open language models with PyTorch and CUDA preinstalled — on a GPU nobody else is using.
Supported creative AI tools
Root access on a Linux machine means you can install these and more. Run them from a notebook or a terminal; web interfaces are reached through an SSH tunnel.
ComfyUI
Node-based Stable Diffusion workflows. Install it on the machine and open its interface through an SSH tunnel.
Automatic1111
A Stable Diffusion web interface. Install it yourself and reach it through an SSH tunnel.
llama.cpp
Run GGUF language models on the GPU from the command line or its built-in server.
Fooocus
Simplified Stable Diffusion, good for beginners. Runs as a web app you open through an SSH tunnel.
InvokeAI
A Stable Diffusion toolkit with a polished interface, installed and opened the same way.
LoRA Training
Fine-tune Stable Diffusion models on your own images, from a notebook or a script.
Ollama
Pull and run open language models with one command.
KoboldCpp
Run GGUF models with GPU offloading for text and story generation.
VRAM requirements by model
Choose the right creative tier based on what you want to generate.
| Model | VRAM needed | Sketch (16 GB) | Canvas (24 GB) | Studio (48 GB) | Masterwork (80 GB) |
|---|---|---|---|---|---|
| Stable Diffusion XL | 8-12 GB | Yes | Yes | Yes | Yes |
| Flux.1 (dev/schnell) | 12-24 GB | Tight | Yes | Yes | Yes |
| Llama 3 8B (Q4) | ~6 GB | Yes | Yes | Yes | Yes |
| Llama 3 70B (Q4) | ~40 GB | No | No | Yes | Yes |
| Mixtral 8x7B (Q4) | ~26 GB | No | No | Yes | Yes |
| SDXL LoRA training | 16-24 GB | Tight | Yes | Yes | Yes |
Why a cloud GPU for AI art?
Nothing to install locally
PyTorch and CUDA come preinstalled, and Studio runs in your browser. Your own computer only needs a browser.
Full NVIDIA driver support
PixelForge workstations come with NVIDIA drivers pre-installed. CUDA and cuDNN work out of the box. No driver conflicts.
A whole GPU to yourself
Your machine's GPU isn't shared with anyone else, and you keep root access to set it up the way you like.
From the collection
Tiers that suit AI art
Canvas
RTX 4090 / 24 GB
$1.22/hr
Great for Stable Diffusion and smaller LLMs
Studio
L40S / 48 GB
$2.47/hr
Faster inference, more VRAM than Canvas
Masterwork
A100 80GB / 80 GB
$4.97/hr
Run 70B+ models and large batch jobs