Skip to main content

pricing

Pay for output, not for access

No subscriptions and no seats — a prepaid wallet, per-unit rates published on every model card, and the price on the button before anything renders. This page explains the model; the live rates are catalog data on the Model Hub.

the meters

One meter per kind of output

Every model bills in the unit its output is measured in. The unit never changes; only the per-unit rate differs between engines.

Video — per second

Every video engine sets its own per-second rate, published on its Model Hub card.

The rate also depends on the resolution and on whether you supply a reference clip, so the card lists the floor (“from”) and the Generate button prices the exact shot: rate × duration, before anything renders.

Images — per picture

Stills, thumbnails and image-to-video source frames bill per generated picture.

Iterate on prompts freely: each picture is a small, known amount, not a subscription decision.

Speech — per audio minute

Voice cloning, text-to-speech and transcription bill by audio minute.

Narration and caption costs scale with runtime, so a longer script never surprises you.

Language — per million tokens

The assistant, script writing and vision models bill per million tokens.

Language models are metered the way language models are everywhere — the published per-million rate is on each card in the Model Hub, next to the video and image rates.

at a glance

What the meter reads

Rates are shown per model on the public Model Hub — compare engines before you even sign up.

Video
per generated second, rate per engine
Images
per picture
Speech & transcription
per audio minute
Language & vision models
per million tokens

how billing behaves

Designed so you never have to ask

If you can't answer “what did that render cost?” without filing a ticket, the pricing has failed. These rules are the whole system.

Priced before it runs

The cost appears on the Generate button and in the assistant's approval card.

Nothing renders until you've seen the number, and each generation in a multi-step assistant plan is priced before it executes. Per-second engines quote the exact charge; a few proxied engines bill on the provider's reported usage, so they show an estimate and settle afterwards — the model card says which one you're on.

A wallet, not a plan

Top up when you want to create; spend it at your own pace.

There is no monthly fee, no seat count and nothing that auto-renews. An empty wallet is a stop, never an overdraft.

Free starter credit

Verify your email after signing up and a starter credit lands in your wallet.

It spends like any balance — enough to try generation, avatars and the editor before you top up.

Refer & Earn

Invite a friend and earn wallet credit when they start creating.

Your referral link, its status and the rewards it has earned live in the Refer & Earn view of your account.

The rate table is the Model Hub

Every engine's per-unit price is live catalog data, public before you sign up. Your wallet balance and per-render history live in your account.