The Vixlens API

3. Create a generation

One endpoint per type. The body takes a model id from the table below, a prompt, and that model's own settings (its model page lists them).

curl -X POST "$VIXLENS_API_BASE/generations/text-to-image" \
  -H "Authorization: Bearer sk_live_..." \
  -H "apikey: $VIXLENS_PUBLIC_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-image-2",
    "prompt": "A matte ceramic mug on a wooden table, morning window light",
    "size": "1024x1024",
    "quality": "medium"
  }'

HTTP/1.1 201 Created
{
  "job_id": "7f3c…",
  "estimated_time_seconds": 3,
  "status": "pending",
  "model": "openai/gpt-image-2",
  "costCredits": …,
  …
}

4. Poll for the result

Generations run asynchronously. Poll the job until its status is completed or failed. Result links are signed and expire, so download the file rather than storing the URL.

curl "$VIXLENS_API_BASE/generations/7f3c…" \
  -H "Authorization: Bearer sk_live_..." \
  -H "apikey: $VIXLENS_PUBLIC_KEY"

{ "status": "completed", "resultUrl": "https://…", "thumbnailUrl": "https://…", … }