How imagen.sh works
One focused interface for generating and editing images across multiple providers, with every request routed through Vercel AI Gateway.
Generate and edit
Start with a text prompt or upload up to two reference images. imagen.sh sends the prompt, inputs, and only the settings supported by the selected model. Completed outputs stay in generation history and can be downloaded, copied, opened, or reused as the next input.
Current models
The model catalog is intentionally explicit so that names such as Nano Banana describe individual Google models, not the product itself.
OpenAI
- GPT Image 2
- GPT Image 1.5
xAI
- Grok Imagine
- Nano Banana 2
- Nano Banana 2 Lite
- Nano Banana Pro
- Nano Banana
FLUX
- FLUX.2 Max
- FLUX.2 Pro
Recraft
- Recraft V4.1 Pro
ByteDance
- Seedream 5.0 Pro
- Seedream 5.0 Lite
- Seedream 4.5
Model-aware controls
Aspect ratio, quality, resolution, and search grounding appear only when the active model supports them. A small AI Gateway classifier infers transparency from the prompt. The selected model always generates first; when it cannot return alpha, imagen.sh runs a second GPT Image step to produce the transparent PNG.
AI Gateway and durable workflows
Users sign in with Vercel, and all generation calls use Vercel AI Gateway rather than separate provider integrations. Vercel Workflow keeps multi-step generations resumable, while Vercel Blob and Supabase store outputs and history.