SDAI — AI Image generator

Dmitriy Moroz

Contains ads

About this app

Create AI images on mobile with SDAI.

SDAI is an open-source AI image generator for Android that gives you a clean mobile workflow for prompting, tuning generation settings, reviewing results, and choosing the AI provider that fits your setup.

Use SDAI when you want a flexible AI art generator instead of a single locked cloud service. Connect to your own AUTOMATIC1111 or SwarmUI server, use AI Horde community workers, call hosted providers such as Hugging Face, OpenAI Images, or Stability AI, or try supported Android local diffusion engines where available.

What you can do:
- Create images from text prompts
- Use positive and negative prompts where supported
- Tune size, seed, steps, CFG scale, samplers, batch options, and provider-specific settings
- Work with Img2Img, Inpaint, LoRA, embeddings, hypernetworks, and model selection when your provider supports them
- Save generated images in a built-in gallery
- Open generation details, reuse prompts, share images, save to device, or export your gallery
- Switch between light and dark themes, languages, and gallery layouts

Why creators use SDAI:
- Bring your own AI provider
- Keep a mobile-first workflow for serious image generation
- Use self-hosted, cloud, community, or local options
- No ads, no telemetry, no provider lock-in
- Open-source and community-driven

SDAI is made for artists, hobbyists, prompt engineers, and builders who want image generation on the go without giving up control. Start with demo mode, connect the provider you already trust, and turn ideas into images from your phone.

Note: feature availability depends on the selected AI provider and app flavor. Some local generation features are Android-only and may not be available in every distribution channel.

Ratings and reviews

Works well, the local mode is particularly impressive. Image generation takes a long time on my Pixel 6 Pro, but then it took a good minute on my desktop computer with a Quadro M4000 graphics card, so this is normal! It's an incredibly cool technology and this app is well made and does the job - the low ratings feel undeserved as I cannot fault this app at all!

If the developers set out to make text entry as difficult as possible they deserve congratulations for their sterling efforts. If, however, you're looking for an interface to your local server which is easier to use than just opening Automatic1111 on a mobile browser theae aren't the droids you're looking for.

The app works good, but to achieve this you need to carefully configure automatic1111 to run with flags --api --listen, and disable Windows Firewall. As for the app itself, I wish there were some more features for saving prompts, and picking them in the txt2img or img2img form.

What’s new

- Added ArliAI provider and improved provider selection with search, filters, and readiness labels.
- Added local SDXL generation with stable-diffusion.cpp.
- Added device benchmark recommendations for local generation.
- Added storage and network usage screens.
- Improved onboarding, gallery performance, localization, and stability.