Pruna AI · Images · standard
P-Image
Half a credit, JPEG out. The other volume tier.
What it costs
| Resolution · 1 MP | 0.5 credits | ≈ $0.01 |
Dollar figures are approximate, at the Starter plan's rate of $19.00 for 800 credits.
What AI video costs per second: 6 findings
What it takes
| Aspect ratios | 1:1 4:3 3:4 16:9 9:16 3:2 2:3 |
| Reference images | Not accepted |
When to reach for it
Pruna's image model, priced with FLUX Schnell at the bottom of the catalog: half a credit, 1 megapixel, one setting, no reference images. It is the second option for the same job: making a lot of images cheaply, where a spread matters more than any one frame.
It differs from Schnell in two ways worth knowing. It returns JPEG rather than PNG, because the provider offers no output-format control, so it is not the model to use when you need transparency or a lossless master. And it offers seven aspect ratios to Schnell's ten, without the 21:9.
Having two models at this price is not redundancy. They render differently, and when a cheap tier keeps missing the mark in a particular way, switching to the other one is a faster fix than climbing the ladder. Neither takes a reference, so neither can hold a subject; that limitation is what the price buys, on both.
How to prompt P-Image
Describe what you want in your own words and Kuro writes the prompt for you. If you would rather write it yourself, this is what P-Image responds to.
Fast text to image and nothing else: the prompt is the entire input, so write everything the picture must show.
It is built for volume — mood boards, thumbnails, background plates, many variations of a layout before you commit. Keep prompts short and change one thing at a time across a batch.
It is at its best on simple, clean compositions and flat subjects.
What it can't do
- No reference images. A saved Subject contributes its written description only, so nothing carries a face or a set from one picture to the next — used for a film's stills, every frame is words alone.
- Fixed at about one megapixel, usually delivered as JPEG.
- Faces, hands and fine texture lose quality quickly.
- Seven aspect ratios: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2 and 2:3.
- Editing one of its pictures moves the edit to a model that can read images.
Use something else when…
- The pictures must look like a specific person or product, or stay consistent across a sequence: Seedream 5 Lite.
- It needs legible lettering: GPT Image 2.5 Flare.
- You need a vector file: Recraft V4.1 SVG.
Guides that cover it
- Which image model to reach for, and when
Twelve image models, from half a credit to twenty-one. What each one is for, where the resolution setting changes the price, and when the cheap tier is right.
P-Image is one pick in the composer.
Every model in this catalog runs from the same prompt bar, priced in the same credits.
Sign up now