One prompt. Every model. Side by side.
Paste an idea — or an image — and see what today’s image generators do with it, together. Compare outputs, keep the prompts and provenance with the pictures, and iterate without hopping between tabs.
- Side-by-side generation — one prompt fans out across the providers you care about
- Web UI — paste or drop an image, write a prompt, pick targets, watch results appear as each provider finishes
- Edit & reference — use an attached image as edit source, image-to-image input, or style/reference
- Prompt craft — randomize, stylize, or rewrite with Claude before you generate
- Round-trip — describe an image, then regenerate from that caption across providers
- Grok video — turn any result into a short video follow-up
- Full archive — prompts, costs, timings, and file provenance stay with the work
| Provider | What you can do |
|---|---|
| OpenAI gpt-image-2 | Text-to-image and image edit; multiple images per run; high detail sizes |
| OpenAI gpt-image-1 / mini | Text-to-image; multiple images per run |
| Ideogram 4.0 | Text-to-image and prompt-guided Remix from an attached image; strong in-image text |
| Ideogram 3.0 | Text-to-image and style/reference from an attached image |
| Black Forest Labs (Flux) | Text-to-image; optional style/reference from an attached image |
| Krea 2 | Krea's own Turbo, Medium, and Large models; text-to-image and style-reference generation |
| Recraft | Text-to-image and image-to-image |
| Google Gemini image | Text-to-image; optional style/reference from an attached image |
| Google Imagen 4 | Text-to-image via Vertex |
| grok-api | Text-to-image and image edit (official xAI API) |
| grok-web | Text-to-image, image edit, and image-to-video via grok.com |
| meta-web | Text-to-image via meta.ai |
| Local ComfyUI | Local Flux Klein and Z-Image targets when enabled |
The primary interface: a local page where you paste from the clipboard, drop a file, or pick a prior input, choose shape and detail, enable the providers you want, and generate.
- Live job cards as each provider finishes
- A text-only badge when an attached image is not sent to a target
- Cost totals as you work
- Input library of previous uploads
- Spell-fix helpers on the prompt
- Make Grok video on any successful result
- Shared-site mode for multi-user use: usernames, day archive, optional login
- Fail closed — missing or ambiguous results are errors; nothing is silently substituted
- Raw saves are verbatim — provider bytes are kept as returned
- Identity preserved — prompts, jobs, and outputs stay correlated end to end
- Compare first — one interesting prompt should show what every current generator would do, annotated and shareable