Simulated AI image API
Build against an image model that behaves like the real thing.
Pixelforge is a fictional image-generation provider. It prices in tokens, works asynchronously, signs its webhooks and fails in all the ways real providers do. No real AI and no real money.
curl https://pixelforge.promptedit.dev/v1/generations \
-H "Authorization: Bearer $PIXELFORGE_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: order-1042" \
-d '{
"prompt": "A lighthouse on a cliff at dawn, oil painting",
"size": "1536x1024",
"quality": "high"
}'Token-based pricing
Text, reference-image, output and reasoning tokens, each at its own rate. Some counts are only known after the image is done.
Async jobs and signed webhooks
Generations are queued and processed in the background. You hear about progress through HMAC-signed webhooks, or by polling.
Realistic failure modes
Rate limits, concurrency limits, overloads, failed generations, retried and duplicated webhooks. Trigger each one on purpose to test.
Expiring image URLs
Finished images are served from signed URLs that stop working after a while, like a real provider CDN.
Here for the PromptEdit assessment?
Pixelforge is the provider you'll integrate with. The brief explains what to build and how it's reviewed.