AI image and video generation
Write one sentence. The studio picks the right model, shows you the price before anything runs, and charges exactly what the generation really cost. Everything it does is also an API.
- Imagesfrom 2,269 toman per imagePhotoreal shots, posters and product photos: add your product (up to 14 reference images) and describe the scene.
- Videofrom 1,731 toman per secondShort clips from a sentence, a still brought to life, or an edit of your own video; some models add sound too.
- Musicfrom 34,622 toman per trackInstrumentals or short songs in the style and mood you describe.
- Voicefrom 215 toman per 1,000 charactersNatural narration of Persian and English text, for audio versions of articles and videos.
From the showcase
Each piece shows the text that made it; press «Make this» to start in the studio with the same text.
The studio in a few minutes
Five short videos: image, video, music, voice, and the automatic model choice with the showcase (Persian subtitles).
1.Generate images in the studio
Product photos, posters or illustrations from a few words; the price before you generate and the exact cost after.
2.Make a short video
A short 720p video from one sentence, with the length and aspect ratio you choose — for your site or Instagram.
3.Create background music
An original instrumental track with Persian or modern instruments, free of copyright worries.
4.Turn text into Persian speech
A natural Persian voice for ads, phone messages or article audio, for a few hundred toman.
5.Auto model choice and the showcase
Auto picks the best model and says why; your creations can appear in the public showcase after review.
Automatic model picking
You do not need to know which model is best at what. A cheap language model reads your request and picks from a table we keep current; the cost of that decision (usually a fraction of a toman) is shown on its own line. Choose a model yourself and your choice always wins.
- A general image
- ByteDance Seed Seedream 4.5
- A photoreal shot
- ByteDance Seed Seedream 4.5
- A product photo
- ByteDance Seed Seedream 4.5
- Anime and illustration
- ByteDance Seed Seedream 4.5
- A poster with text
- Google Nano Banana 2 (Gemini 3.1 Flash Image)
- An edit of your photo
- Google Nano Banana 2 (Gemini 3.1 Flash Image)
- Video from text
- Google Veo 3.1 Fast
- Animate a photo
- Kling Video v3.0 Standard
- A cheap, quick draft
- SpaceXAI Grok Imagine Video
- A full track
- Google Lyria 3 Pro Preview
- A short clip or jingle
- Google Lyria 3 Clip Preview
- Read English text
- MiniMax Speech 2.8 HD
- Read Persian text
- Google Gemini 3.8 Flash TTS
Money and privacy, plainly
- Price before you generate
- You see a toman estimate before every run; afterwards only the real cost is charged.
- Failed runs are free
- If a run fails or the model maker refuses it, nothing is charged.
- No credit packs
- No subscription and no bundle to buy up front; it comes out of the same uttapen toman wallet.
- Private unless you say so
- Your outputs are yours and kept for 30 days. The public showcase only shows items you opted in and we approved.
The same thing, over the API
If you write code, the same sk-up- key and the official OpenAI SDK are enough. Leave the model empty, or send uttapen/auto-image, and automatic picking does the rest.
from openai import OpenAI
client = OpenAI(base_url="https://api.uttapen.ir/v1", api_key="sk-up-...")
img = client.images.generate(
model="uttapen/auto-image",
prompt="پوستر یک کافه با متن «قهوهٔ صبح» روی دیوار آجری",
)
print(img.data[0].url) # a signed link, valid for 24 hoursFrequently asked
- Which models make the images and videos?
- Models from the leading makers — Google (Veo, Nano Banana, Lyria), ByteDance (Seedream, Seedance), Kling, MiniMax and dozens more. The full list, with each model's toman price, is on the models page.
- How is it priced?
- Video per second (depending on resolution and sound), images per image, voice per character and music per track. You see an estimate first and an amount is held; after the run only the real cost is charged and the rest returns.
- What if a run fails?
- Nothing is charged and the hold is released in full — whether the failure is ours or the model maker refused the request.
- What will it not make?
- Sexual content, anything harmful involving children, graphic violence, and images or voices made to look like real people. Those requests are refused before anything is charged.
- Can anyone see what I make?
- No. Outputs are private and kept for 30 days so you can download them. Only if you tick «show in the showcase» and we approve does an item appear publicly.
- Can I use it from my own app?
- Yes. All of it is available through an OpenAI-compatible API: image generation, asynchronous video with status polling, text to speech, and music.