A small native Mac app that turns a prompt into a batch of images.
- Free with your ChatGPT subscription. The app reuses your local
codex login, so there is nothing new to buy. - Works with any OpenAI-Images-compatible relay too. Bring your own key and see the per-image price before you generate. Tested with
gpt-imageand thenano-bananafamily. - 2.4 MB of SwiftUI, zero third-party dependencies.
- Each image is its own request. Submit the next batch while the previous one is still running.
- Results land in a folder you choose, and that folder is the history. Quick Look, drag to Finder, no database.
- Size options match what each backend actually accepts. We verified against live endpoints and removed the ones that silently fall back.
Requires macOS 15+ and Xcode 16+.
git clone https://github.com/Suge8/Image-Studio.git && cd Image-Studio
make install && make runPick one or both. Switch anytime from the capsule at the top left.
| Channel | Setup |
|---|---|
| Codex | Run codex login once in Terminal and choose ChatGPT. |
| Relay | Settings → Third-party Relay → base URL and API key → Save & Verify. Keys go to the macOS Keychain. |
Type a prompt, press ⌘↩. Finished images drop into the gallery one by one.
| Iterate on a result | Right-click → Use as Reference |
| Add reference images | Drag, paste (⌘V), or click the dropzone, up to 16 |
| Reuse a prompt | ★ favorites (a logo-board template ships built in), clock icon for history |
| Preview | Select an image, press Space |
| One slot failed | Hover it and retry only that one |
make test # unit tests
make package # Release zip → dist/Architecture and design docs live in docs/. Start at AGENTS.md. Contributions welcome: CONTRIBUTING.md · SECURITY.md · Apache-2.0