Imference Desktop generate on your own GPU, or in the cloud. One app.
Run SDXL and Z-Image locally — private and free on your hardware — or generate through imference.com with credits or x402. No subscription, no lock-in.
Mode
Local
Your GPU, your images
Mode
Cloud
No GPU required
Models
SDXL · Z-Image
+ full cloud catalog
Prompt
cinematic portrait, soft rim light, 85mm lens…
Two ways to generate
Switch between local and cloud at any time — same interface, same gallery.
Local — your GPU
Private, unlimited, free to run
One-click engine install, then generate with SDXL and Z-Image directly on your machine. Images never leave your computer.
- NVIDIA GPU (CUDA) on Windows · Apple Silicon on macOS
- Isolated environment — nothing touches your system Python
- Models downloaded on demand (~6–7 GB each)
Cloud — imference.com
No GPU required, full model catalog
Generate on imference.com servers straight from the app. Pay with API credits or x402 (USDC on Base — no account needed).
- Works on any machine — laptops, no GPU, low VRAM
- Full hosted catalog: realism, anime, video models
- Same prompts, params and gallery as local mode
Built for actually making images
Not a wrapper around a text box — real controls, real history.
Per-model parameters
Format, steps, CFG, seed, quality tags, negative prompt — with sensible defaults pulled from the model catalog.
Image-to-image
Start from an existing image and steer it with a prompt — locally or in the cloud.
Metadata-rich gallery
Every generation is saved with its prompt, model and params. Filter, search, and reopen in a fullscreen viewer.
Heads-up on first launch
The app isn't code-signed yet, so your OS will show a warning the first time you run it. That's expected — signing and in-app auto-update are on the way.
- Windows: SmartScreen → More info → Run anyway.
-
macOS:
right-click the app → Open (or
xattr -cr "/Applications/Imference Desktop.app").
Every release ships a checksums.txt so you can verify your download.
Requirements
System
Windows 10/11 (x64) or macOS 12+
Local generation
NVIDIA GPU (CUDA) on Windows, or Apple Silicon on macOS
~6–7 GB of disk per model, downloaded on demand
Cloud generation
An imference.com API key, or a funded x402 wallet
Start generating in minutes
Download, launch, pick local or cloud — that's it. Free and open on GitHub.