Your creations stay yours.
While other AI tools send your prompts to remote servers, Conjure generates everything locally on Apple's Neural Engine.
No Cloud
Every image is generated on-device. Nothing is uploaded. Ever. Works completely offline.
No Accounts
Open the app and start creating. No sign-up, no login, no email required.
No Tracking
Zero analytics. Zero telemetry. We literally don't know what you create.
Everything you need.
Professional-grade Stable Diffusion that fits in your pocket.
Hardware-accelerated on Apple silicon. Generate images in ~4 seconds.
Cinematic, Anime, Fantasy Art, Nature, People, Enhance, or go freestyle.
All creations saved locally with full metadata. Export to Photos anytime.
Bring your own Stable Diffusion models. Not locked to one provider.
Generate anywhere — airplane, subway, the middle of nowhere. No WiFi needed.
iPhone and iPad. Same app, optimized for each device.
See what Conjure creates.
Real outputs from the app. Create, browse, and save — all on device.
Three steps. That's it.
Describe
Type what you want to see in plain language.
Choose style
Pick from 7 creative presets or go freestyle.
Generate
Watch your image appear in seconds. Save or share.
Powered by Stable Diffusion XL.
Real AI. Running locally on your iPhone.
Conjure runs DreamShaper XL Lightning, a Stable Diffusion XL (SDXL) model optimized for Apple's Neural Engine. SDXL generates 1024×1024 images — the same architecture used by professional AI art tools, running entirely on your device.
To fit a full SDXL model on a phone, we use 4.77-bit mixed-bit palettization — a custom quantization technique that assigns different bit depths to each of the model's 792 weight tensors. Critical layers keep higher precision. Less sensitive layers compress further. The result: desktop-quality Stable Diffusion in your pocket.
Unlike cloud-based AI image generators, Conjure never sends your prompts or images to a server. Every step of the diffusion process — from text encoding to latent denoising to VAE decoding — runs locally on Apple silicon.
Frequently asked questions.
What AI model does Conjure use?
Conjure uses Stable Diffusion XL (SDXL) via DreamShaper XL Lightning, a distilled model optimized for fast generation. It produces 1024×1024 images in approximately 4 seconds on iPhone.
Can I generate AI images on my iPhone without internet?
Yes. Conjure runs entirely on-device using Apple's Neural Engine. Once installed, it works completely offline — no WiFi, no cellular data, no cloud connection required.
Is Conjure really private?
Conjure has zero analytics, zero telemetry, and requires no account. Your prompts and generated images never leave your device. We have no servers that receive user data.
What devices does Conjure support?
Conjure runs on iPhone and iPad with Apple's Neural Engine (A-series and M-series chips). It requires iOS/iPadOS 17.0 or later.
How does Stable Diffusion run on a phone?
Through mixed-bit palettization — a quantization technique that compresses the model's 792 weight tensors to an average of 4.77 bits per weight, fitting the full SDXL pipeline in device memory while preserving image quality.
Is Conjure free?
Conjure is available on the App Store for iPhone and iPad.