Frontier Multimodal AI. Made Affordable.

Image, video, audio and lipsync models on a stack we run end to end. One API, a playground, pay as you go.

Browse the models Request access

Glow serum
AI UGCAI avatar
Ines in NYC
Image to videoFrom one still
Ines on bass
Consistent charactersAI avatar

What you can make

Each tile is one kind of output our models produce. Open one for real examples and the model behind them.

Service

Ad creation

Send us a product page. Finished video ads come back in the formats your channels take.

Service

Consistent characters

See what our system can make.

Real outputs from our models. Pick a category to filter the reel, or click any clip to watch it full size.

VideoPodcast-style ad, AI presenter
VideoProduct ad, AI presenter
AvatarAI avatar, the same face every cut
AudioVoice and score for an ad0:15
VideoImage to video, AI presenter
VideoFirst and last frame: an empty room fills with furniture
VideoFirst and last frame: a peony bud opens
VideoShort film scene
VideoShort film: Dunewake, a cut-down of the 93-second film
ImageGenerated street still
AvatarConsistent character: the same presenter in nine scenes
VideoFashion ad, AI presenter
VideoTry-on, before and after
VideoTwo AI presenters, one scene
VideoSkincare ad, podcast style
VideoFilm shot: weightless in orbit
Video3D: a bear in a flying saucer

How it works.

Three steps, the same for every model.

01
Pick a model
Browse the catalog, read the price per unit, and try the model in the playground before you write a line of code.

Request

POST https://api.strontium.ai/jobs/async/{slug}
Authorization: Bearer <BASE64_TOKEN>
{ "modelParams": { "prompt": "A lighthouse at sunset" },
  "webhookUrl": "https://your-server.com/webhook" }

201 Created  { "jobId": "abc123" }

Send the request

One POST from your code with the model's inputs, or a form in the playground. Each job is billed from your prepaid balance.

Webhook

POST https://your-server.com/webhook
{ "id": "abc123",
  "succeededAt": "2025-02-12T10:31:02.512Z",
  "unitsUsed": 5,
  "outputPath": "jobs/abc123/output.mp4",
  "pricePerUnit": 0.02 }

Get the output

The result lands on your webhook or in your dashboard, with the job's status, the units used and what it cost.

From a product photo to a finished clip.

One photo goes in. The script, the voice, the presenter and the cut are generated. Watch the steps run.

Your photo

The Hugease koala, a grey plush toy with a softly glowing belly, photographed on white

One product shot.No studio, no presenter, no script yet.

What happens

1Reads the productThe photo and the product page: what it is, who it is for.
2Writes the script, records the voiceA hook and a script for the channel, spoken by a generated voice.
3Films the presenterA generated presenter delivers it with the product in hand.
4Lipsync, cut, three sizesMouth matched to the voice, cut to length, delivered 9:16, 1:1 and 16:9.

Finished clip

Hugease product story.15 s, vertical, made from that one photo.

A replay of the steps. Nothing is generated on this page.

What our clients say

Working with them completely changed the way Mr.Fris grows. For the first time I am able to scale on Meta consistently instead of guessing what works.
The biggest win is that I am able to scale across multiple EU markets at once. Their creative and media buying let us move into new countries without the usual risk, and the numbers held up in every single one.
Because of them I am able to grow my business, double my business, and keep expanding across the whole of Europe. I could not have scaled like this on my own.
KlaasFounder, Mr.Fris
Their AI native videos changed how we go to market. Instead of slow, expensive shoots we finally had scroll stopping creative we could ship fast and test at real volume.
That creative did the heavy lifting on our growth. We drove more than 10,000 mobile app installs in our region on the back of their videos, at a cost per install we could actually scale on.
They understood our product and our market and turned it into ads that made people download. For an app like ours that is exactly the partner we needed.
KoenFounder, Tinzo
Before them, every ad creative meant booking a full production shoot. Models, photographers, studio time, thousands of euros gone before we even knew if the creative would work.
They replaced all of that. We now get scroll stopping, on brand creatives that look like a luxury campaign, without a single traditional shoot.
They have saved us thousands per shoot across all our ad creatives, and we can test far more angles than we ever could before. For a jewelry brand where the visuals are everything, that changed the game.
Chouaib TamniApollonia
For a long time our advertising was bottlenecked by creators. Every batch of content meant long delays and endless back and forth, and we would wait weeks before we could even test a new angle.
AI UGC changed the game for us completely. We can now produce ad creative on demand, at a fraction of the cost, and the ads actually perform. Speed, efficacy, and cost all moved in our favour at the same time.
It finally let us ramp up our advertising instead of being held back by production. We ship more, test more, and scale faster than we ever could before.
Hishin RegenBlood sugar support supplement
Since we started using AI native videos, our engagement across social has gone through the roof. For the first time we can actually show what makes our product special instead of just talking about it.
The videos let us showcase the real benefits of Mmelt in a way that stops the scroll. People do not just watch, they react, comment, and share.
It gave us a content engine we can keep feeding, and the jump in engagement has been the clearest signal that it is working.
MmeltBetter-for-you chocolate and candy