Get a key

Pricing

Priced per finished result.

Images from $0.03, video from $0.18 a clip, and the exact charge on every response.

$0.03an image
$5to start
$0.00for a failure
Estimated monthly$146.401,200 images and 240 seconds of 1080p video, at PremiumChange the volume

Every price, one table

What each tier costs

  • Image: billed per finished image, whatever the router picks inside the tier.
  • Video: billed per second of the completed clip. Air and Lite cap at 720p, Premium and Pro at 1080p, Max Pro at 4K.
  • Text: OpenAI-compatible and billed per token. The figures are a typical response.
TierImageper imageVideoper clipTextper responseVoiceper 1,000 characters
AirHigh volume, the simplest work.$0.03$0.03 to $0.05$0.25$0.18 to $0.44$0.03
LiteEveryday work, stronger quality.$0.06$0.03 to $0.10$0.29$0.24 to $0.33$0.03
PremiumProfessional content.$0.07$0.04 to $0.10$0.52$0.44 to $0.86$0.04$0.03 to $0.04
ProComplex commercial work.$0.08$0.05 to $0.10$0.84$0.28 to $1.31$0.05
Max ProMaximum fidelity.$0.14$0.06 to $0.25$1.85$0.81 to $3.43$0.08$0.06 to $0.10

Voice is $0.09 per 1,000 characters of input text at every tier. Started thousands count in full.

Live status is not printed here, because a printed badge goes stale. Every tier publishes its own, and GET /v1/status returns one line per tier.

corent.tech/status

Preferences

Or say how to optimize

Pass one word instead of a tier. It resolves like this.

PreferenceOptimizes forImage tierVideo tier
cheapCost firstAirAir
fastLatency firstAirAir
balancedCost, speed and quality equallyLitePremium
qualityOutput quality firstPremiumPro

The router, x-rayed

Watch one prompt get routed

The profile and the scores below are read out of the text you type.

Try

Prompt profile

  • task image
  • graphic and type
  • needs legible text
  • no human subject
  • complexity 60%

route image-premium

Read from the prompt on every request. Edit the line above and watch it move.

Candidates, scored for this prompt

  1. A9.73routed
  2. B8.61below floor
  3. C9.48held
  4. D9.17held
  5. E8.60below floor
  6. F8.82below floor

Legible type is required, so the router weights typography over raw fidelity. A clears the Premium floor at 9.73 and takes the route. 3 fall under it.

Estimate

What a month costs

Typical prices, multiplied by your volume. The arithmetic is on the receipt.

Images per month
1,200 images
Video per month
240 seconds
Tier
Video resolution

The tier is the same one the x-ray is routing on.

Billing

Prepaid credits, no subscription

  1. 01

    Top up

    From $5 by card or $10 in crypto. It lands once the payment confirms.

  2. 02

    Generate

    Each request checks the balance, runs, and debits the exact cost after it completes.

  3. 03

    See the receipt

    The charge comes back on the response and as a line in your dashboard.

Failed generations are never charged.

You pay for finished results, not attempts.

The business case

What it replaces.

Vendors
One agreement, not one per lab.
Billing
One prepaid balance, one statement.
Failures
Never billed, so nothing is spent on work you did not receive.
Commitment
No lab lock-in, the router moves when a better model ships.

FAQ

Pricing questions

How does pricing work?

Pay as you go on prepaid credits. Add credit from $5 by card or $10 in crypto, then each generation debits the exact cost of that result. No subscription, no monthly fee.

How is video billed?

Per second of output, on the completed clip. A longer clip costs more at the same tier, and the exact charge comes back on every response.

Do I pay for failed generations?

No. The balance is debited only after a result completes. Corent retries across models first; if everything fails, the job is marked failed and you are charged nothing.

Do I choose the model, or does Corent?

Either. Pass a tier or a one-word preference and the router picks and reroutes. Or pin an exact model by name, billed flat cost-plus and never substituted.

Who owns what I generate?

Not documented on this site. Get the licence terms in writing before you buy, rather than taking them from a marketing page.

Ask us: support@corent.tech
Do you train on my prompts?

Not documented on this site, so treat it as unanswered. Get it in writing if your prompts carry anything confidential.

Ask us: support@corent.tech
Can I get an invoice, or pay by bank transfer?

Charges are itemised: the exact cost returns on every response, and /v1/account/usage totals recent spend. Top-up is card or crypto only, and invoicing is not documented here.

Ask us: support@corent.tech
What happens to unused credit?

The balance is prepaid and only ever debited by a finished result, so nothing draws it down on its own. Expiry and refunds are not documented here.

Ask us: support@corent.tech
Is there support?

Email support, and watch live per-tier status at corent.tech/status. Response times and any paid support tier are not published here.

Ask us: support@corent.tech

Before you buy

What you are signing up to.

Uptime
Live per-tier status is published rather than printed on this page, and a failing model is rerouted around in about 2ms.corent.tech/status
Spend
Every key carries a rate limit, a concurrent-job limit and a daily cost cap. New keys start at $10 a day, and raising a cap needs a dashboard login.
Billing
Prepaid and drawn down in cents. The exact charge returns on every response and as a line in the dashboard, and a failed generation reads billed: false.
Support
One address, answered by the people who run the router. Ask before you buy, not after.support@corent.tech

Data handling, retention and ownership of output are not documented anywhere on this site. Ask for them in writing before you commit.