DashboardAPI
Documentation

Image Generation

POST /v1/images/generations handles image generation, image editing, and — for some models — video generation. The same endpoint covers all three; the model you pick determines the behavior.

Synchronous example

By default, image models return the result inline. Use this when the model is fast (Flux, DALL·E, Imagen, gpt-image-1.5, etc.) and your client can wait for the response.

1curl -X POST https://api.navy/v1/images/generations \
2  -H "Authorization: Bearer sk-navy-YOUR_KEY" \
3  -H "Content-Type: application/json" \
4  -d '{
5    "model": "flux",
6    "prompt": "A naval command room at dusk, cinematic lighting",
7    "size": "1024x1024"
8  }'

Asynchronous example with sync false

Force a job-based workflow with "sync": false. The endpoint returns a job ID immediately and you poll for the result. Use this when your client can't hold a long-lived HTTP connection (serverless functions, browser fetches, gateways with short timeouts) or when you want to queue background work. See the Job Polling page for the full flow.

1curl -X POST https://api.navy/v1/images/generations \
2  -H "Authorization: Bearer sk-navy-YOUR_KEY" \
3  -H "Content-Type: application/json" \
4  -d '{
5    "model": "flux",
6    "prompt": "A naval command room at dusk, cinematic lighting",
7    "size": "1024x1024",
8    "sync": false
9  }'

Parameters

  • model (string, required) — Image or video model ID such as flux, dall-e-3, gpt-image-1.5, imagen-4, veo-3.1, sora-2
  • prompt (string, required) — Text description of the desired image or video
  • size (string, optional) — "1024x1024" or aspect ratio like "16:9"
  • aspect_ratio (string, optional) — Alternative to size for some models (e.g. "1:1", "9:16")
  • image_url (string or array of up to 5 strings, optional) — Input or reference image(s) for editing. Accepts HTTPS URLs or data: URIs. Use an array to combine multiple references on supported models.
  • negative_prompt (string, optional) — What to avoid in the generated image
  • seed (integer, optional) — Random seed for reproducible generation
  • quality (string, optional) — "low" or "medium" (other values are stripped)
  • style (string, optional) — DALL·E 3 only: "vivid" or "natural"
  • seconds (number, optional) — Video duration in seconds, clamped to 0–10 (video models only)
  • sync (boolean, optional) — Defaults to true for image models. Set false to get a job ID and poll asynchronously. Video models always run async regardless of this flag.
  • response_format (string, optional) — "url" (default) or "b64_json" to receive a base64-encoded payload. Accepts "base64" and "base64_json" as aliases.

Async behavior at a glance

  • Image models, default (`sync: true`) — Response returns inline with data[].url or data[].b64_json.
  • Image models, `sync: false` — Response returns a job; poll GET /v1/images/generations/:id.
  • Video models — Always async by default. You almost always receive a job; expect generation times up to 10 minutes.

Notes

  • Compatible editing models accept image_url as a string or an array (max 5).
  • Multi-reference editing example: "image_url": ["https://example.com/a.png", "https://example.com/b.png", "https://example.com/c.png"]
  • Responses with b64_json may also include a url field pointing to a hosted copy so you don't have to upload it yourself.
  • n (multiple images per request) is not supported — request one image per call.
  • See the Job Polling page for the full async lifecycle, status values, and error codes.
← PreviousResponsesNext →Job Polling