Dual-mode generation
It runs text-to-image by default. Upload a reference photo and it switches straight into image-to-image mode, swapping in the 70+ model set built for editing rather than generating from scratch.
One workspace that generates from a blank prompt or edits an uploaded photo, switching model sets automatically — with up to 14 reference images per request and no content filters.
It's the platform's dedicated workspace for still images — one canvas that covers both generating from nothing and editing something you already have.
It runs text-to-image by default. Upload a reference photo and it switches straight into image-to-image mode, swapping in the 70+ model set built for editing rather than generating from scratch.
Feed up to 14 reference images into compatible edit models like Nano Banana 2 Edit. A multi-select picker with order badges keeps character and product consistency across every generation.
Quality and resolution controls only appear for the models that actually support them, so switching between a fast draft model and a 4K-capable one never means hunting through settings that don't apply.
Both categories run 70+ models each; here's a sample of what's inside, from fast local options to the newest hosted releases.
z-image-turbo also runs locally in the desktop app's bundled sd.cpp engine — no API key required.
New releases land in the same picker as everything else, with no separate signup or waitlist. The most recent additions are Nano Banana 2 for generation and its companion Nano Banana 2 Edit for reference-based edits, plus Seedream 5.0 and Seedream 5.0 Edit for the same dual-mode pairing, alongside MiniMax Image 01. GPT-4o Image, Ideogram v3 and Midjourney v7 are already in the same list.
The same flow whether you're starting from a blank prompt or editing a photo you already have.
Type a prompt for text-to-image, or drop in a reference photo to switch straight into image-to-image mode.
Choose from 70+ text-to-image or 70+ image-to-image models; resolution and quality options appear automatically for models that support them.
Download the result, or send it straight into Video Studio as a first frame, or Lip Sync Studio as a portrait.
This is one of 14 purpose-built workspaces on the platform. Here's where to go next.
Turn a still into motion, or generate a clip straight from text — 85+ text-to-video and 120+ image-to-video models.
Generate and edit AI audio and music from a text prompt.
Animate a portrait you made here, or re-sync lips on an existing video — 9 dedicated models.
Photorealistic shots with pro camera, lens, focal length and aperture controls.
Chain this studio's output into video and audio models as one repeatable, node-based pipeline.
70+ text-to-image models, 70+ image-to-image models, up to 14 reference images. No subscription, no account required to try it.
Try Free