POST /v1/images/generations
API Reference
POST /v1/images/generations
Image generation with any model from the catalog
POST
POST /v1/images/generations
Request body
The namespaced model ID, e.g.
fal/flux-1.1-pro.Image description. Maximum 4000 characters.
Number of images to generate (1–8). Each image bills independently.
Desired size. What’s supported depends on the model:
- OpenAI gpt-image-1:
1024x1024,1536x1024,1024x1536 - OpenAI dall-e-3:
1024x1024,1792x1024,1024x1792 - fal/Recraft/Ideogram:
1024x1024,1024x768,768x1024, etc.
standard | hd for OpenAI. Other models ignore it.Response
The URL is ours (Supabase Storage), not the provider’s. It’s persistent and doesn’t expire. Same security model as fal.media, Cloudinary, or unsigned S3: the URL includes 2 UUIDs (
org_id + request_id) making it unguessable.Examples
Expected latency
| Model | P50 | P95 |
|---|---|---|
| DALL·E 3 | 8s | 20s |
| GPT Image 1 | 15s | 45s |
| Imagen 4 | 6s | 18s |
| Grok 2 Image | 10s | 30s |
| Flux 1.1 Pro | 7s | 25s |
| Flux 1.1 Pro Ultra | 12s | 40s |
| Recraft V3 | 8s | 28s |
| Ideogram V2 | 12s | 35s |
| SD 3.5 Large | 10s | 30s |