Midjourney, by API

Image generation you call from your own code. One prepaid balance, one price per job, and a reference that matches what the service actually does.

How it works

Pay for the jobs you run

$0.07 an image, $0.02 for the cheaper calls. No subscription, no seats, no monthly minimum. Top up in cryptocurrency and spend it when you need it.

A failed job costs nothing

The price is taken when the job starts and put back the moment it fails, before you have finished reading the error. You never have to ask.

Written from measurement

Every limit and behaviour in our reference was measured against the live service. Where we do not know something, it says so rather than guessing.

Start in one request

curl https://omegen.ai/api/v1/generations \
  -H "Authorization: Bearer omg_your_key" \
  -H "Content-Type: application/json" \
  -d '{"action":"imagine","prompt":"a red apple on a marble table --ar 16:9"}'

One job returns four images. Ask for the job again until its status changes: there is no callback to receive, so nothing on your side has to be reachable from the internet. The full reference is four pages and covers every action.

What you can call

Midjourney 8.2

Every parameter the model accepts works here, written the same way: --ar, --style, --sref, --chaos, --no. Name another version with --v and yours is used instead.

Price	Actions
$0.07	`blend` `edit` `high_variation` `imagine` `low_variation` `pan_down` `pan_left` `pan_right` `pan_up` `reroll` `retexture` `variation` `vary_region` `zoom_out_1_5` `zoom_out_2`
$0.02	`describe` `upscale`

Prices on this page are read from the price list the service actually charges from, so what you see here is what you pay. More on how billing works.

What comes next

Midjourney is the first model here, not the only one planned. Video and music go behind the same key, the same balance and the same request shape, so an integration written now keeps working when they arrive.