Systèmes opérationnels

Vrex · L'atelier d'IA multimodale

Chaque moteur.Un seul solde.Meilleur prix.

Les studios voix, image et vidéo que vous mettriez en file séparément — les meilleurs fournisseurs sous un même toit, facturés depuis un solde unique, aux tarifs pay-per-use les plus bas que nous puissions négocier. Pas trois abonnements, pas de verrouillage moteur, pas de palier surprise.

Commencez avec 1 $ offert — sans carte. Ensuite, paiement à l'usage en dollars réels ; sans abonnement, et votre solde n'expire jamais.

Planche I — Voix : plus de 600 langues, le moteur qui colle au script
Planche I — Voix : plus de 600 langues, le moteur qui colle au script
Planche II — Image : chaque générateur de pointe, choisi selon le prompt
Planche II — Image : chaque générateur de pointe, choisi selon le prompt
Planche III — Vidéo : huit secondes de cela, mises en mouvement
Planche III — Vidéo : huit secondes de cela, mises en mouvement

Feuille de spécimen

La même idée,tous les médias.

Un exemple travaillé. Le prompt ci-dessous ne bouge pas ; les studios se relaient. Même solde, même tableau de bord — choisissez le moteur qui convient, ou laissez Vrex choisir pour vous.

Le prompt

Un gardien de phare consigne le dernier lever de soleil avant l'automatisation de la lampe.

Voix — Narrateur buriné, cadence lente, chaleur d'embruns — lu à voix haute en entier.
VoixNarrateur buriné, cadence lente, chaleur d'embruns — lu à voix haute en entier.
Image — Lumière froide de l'aube rasant une tour ; laiton, brume, longue ombre sur la pierre mouillée.
ImageLumière froide de l'aube rasant une tour ; laiton, brume, longue ombre sur la pierre mouillée.
Vidéo — Huit secondes : le faisceau balaie une fois, puis l'ampoule s'éteint.
VidéoHuit secondes : le faisceau balaie une fois, puis l'ampoule s'éteint.
À lire comme une dernière entrée de journal
01 / 04

The Studios

Three rooms. One door key.

Each medium has its own bench. Your balance opens all of them — switch rooms mid-project without buying another pass.

Coral audio waveform on warm paper — a voice sample

The Voice Room

The Voice Room

Text to speech that reads like it was performed, not parsed.

  • $0.005 per 1,000 characters — a full chapter for the price of a coffee sip
  • 600+ languages and accents from the OmniVoice model (sherpa-onnx)
  • Clone a voice from a short sample, then narrate anything in it
Enter the Voice Room
Editorial still life — a single generated image plate

The Image Room

The Image Room

Stills with a point of view — editorial, photoreal, or printed-look.

  • $0.013 per image — roughly 384 frames from a single $5 top-up
  • Reference images in for style, character, and composition control
  • Portrait, landscape, or square; download print-ready files
Enter the Image Room
A single film frame from a generated video shot

The Video Room

The Video Room

Short motion with intent — a shot, not a slideshow.

  • $0.017 per second — eight seconds of footage for about $0.14
  • Text-to-video, image-to-video, or build a shot from ingredients
  • Aspect and duration you set; commercial rights on every render
Enter the Video Room

The Method

How a prompt leaves as finished work.

Four moves from blank page to downloadable file — no pipeline to configure, no rooms to unlock.

  1. Write the brief

    The brief

    A sentence, a mood board in words, or a full script. The same brief feeds every medium.

  2. Pick the room

    The room

    Voice, image, or video. Set the dials that matter — language, aspect, duration — and skip the rest.

  3. Pull the proof

    The proof

    Vrex renders in seconds and bills only the work it ships, charged against your one balance in real dollars.

  4. Take the file

    The file

    Download MP3, WAV, PNG, WebP, or MP4 — commercial rights included, yours to publish.

The Ledger Reads Plainly

Every engine, one balance — instead of three subscriptions.

Most creators stack a voice plan, an image plan, and a video plan: three bills, three logins, three pools you can't cross. Vrex aggregates the top engines under one balance, at pay-per-use rates we keep below market.

Vrex

What you payPay-per-render in real dollars — typically below each engine's own retail
Which engines you getEvery leading voice, image, and video model on one bench
Where the money livesA single pool shared across voice, image, and video
When it expiresNever — unused balance carries forever
Commercial rightsIncluded on every render, every medium

Three subscriptions

What you payThree recurring fees, charged whether you create or not
Which engines you getWhichever model that vendor ships — switching means switching plans
Where the money livesSiloed allowances that can't move between tools
When it expiresMonthly allowances reset; whatever you didn't use vanishes
Commercial rightsVaries by plan tier and add-on

The Price Sheet

Priced by the unit. Paid from one balance.

No tiers to decode, no seats to count. You pay for exactly what you render, in real US dollars, from one pool.

Priced by the unit.
MediumUnitPrice
Voice — text to speechper 1,000 characters$0.005
Image generationper image$0.013
Video generationper second$0.017

Ce que 5 $ achètent

Une recharge de 5 $. Le média de votre choix.

Voix1,000,000 caractères
Images≈ 384 images
Vidéo≈ 294 secondes

Mélangez librement — les mêmes 5 $ s'étirent sur les trois.

Top-up packs

Add balance once. Spend it across every medium, on your own schedule.

Starter
$5
$5 usable

A first run at all three rooms

BasicBest value
$15
$17 usable · +$2 bonus

For a working week of output

Pro
$50
$60 usable · +$10 bonus

For shipping a full project

Balance never expires · Prices shown per unit. Top-up packs add real USD to one shared balance — no subscription, no auto-renew, nothing taken back.

In the Field

Made for people who ship more than one thing.

Most projects need a voice, a picture, and a clip. Here's who runs the whole brief through one balance.

Nº 01 — In the Field
The Podcaster — Records the episode, then needs everything around it — without three more tools.

The Podcaster

Records the episode, then needs everything around it — without three more tools.

  • An intro and outro read by a cloned host voice
  • Square cover art and a quote still per episode
  • A short audiogram clip for the feed
Nº 02 — In the Field
The Marketer — Ships a campaign a week and can't wait on a queue for each asset.

The Marketer

Ships a campaign a week and can't wait on a queue for each asset.

  • Ad voiceover in the launch market's language
  • A set of editorial product stills, one prompt
  • Eight-second motion cutdowns for paid social
Nº 03 — In the Field
The Filmmaker — Boards and pitches fast, then proofs ideas before the real shoot.

The Filmmaker

Boards and pitches fast, then proofs ideas before the real shoot.

  • Scratch narration in the lead character's voice
  • Lookbook frames to lock the palette and lens
  • Animatic shots to test a sequence end to end

Pour mémoire

Aucun battage inventé. Juste les faits sur la planche.

Des chiffres que nous assumons — aucun volume d'usage non mérité, aucun benchmark non exécuté.

600+Langues parlées par le modèle vocal
3Voix, image, vidéo — une seule réserve
0,013 $Par image, en dollars réels
0,017 $ / sPar seconde de vidéo générée
5 $ → 1MCaractères de voix pour une recharge
JamaisLe jour où votre solde expire

Questions Answered

Before you ask at the front desk.

How does pricing actually work?

You pay per unit, in real US dollars: $0.005 per 1,000 characters of speech, $0.013 per image, $0.017 per second of video. No subscription and no seats — your balance is only spent when you render something.

Can I use what I make commercially?

Yes. Commercial rights are included on every render across all three mediums — voice, image, and video. Publish, sell, or broadcast what you create.

Does my balance ever expire?

No. Balance you add never expires and never resets. Top up once, spend it next week or next year — it stays exactly where you left it.

Which languages does the voice support?

Text to speech runs on the OmniVoice model (sherpa-onnx), which supports 600+ languages and accents. Write in your language; it reads in it.

Can I clone a voice?

Yes. Provide a short, clean audio sample and Vrex builds a voice you can then narrate anything in — across languages, on the same balance.

What files do I get out?

Voice exports as MP3 or WAV, images as PNG or WebP, and video as MP4. Standard formats, ready to drop into any editor or player.

Is there a free way to try it?

Yes. Sign up and $1 lands in your balance — about 76 images, 58 seconds of video, or 200,000 characters — enough to run a real prompt through all three rooms before you add a cent. No card required.

Do the three mediums share one account?

They do. One login, one balance, three rooms. Move from a narration to a still to a clip without switching tools or paying another bill.

Dark atelier still — a quiet workbench bathed in low light

Last Plate

Pick the engine. Pay from one balance.

Open the atelier with $1 free — enough to test-drive every studio (about 76 images, 58 seconds of footage, or 200,000 characters of speech). Add five dollars and that's ~384 images, ~294 seconds, or a million characters — at rates we keep below each engine's own retail.

No subscription · No card to start · Balance never expires

Vrex — Synthèse vocale, image et vidéo par IA sur un seul solde