Pricing
A flat monthly subscription over the models we serve direct, with pay-as-you-go for the rest of the catalog. No seats and no minimums: a monthly allowance with a rolling 7-day cap, and every number here is one you can watch move in the console.
One flat price for the models we serve direct, sized for real production traffic rather than a trial.
Models included
Billed monthly through Stripe, renews until cancelled. Credits never buy the subscription. Cancel any time and access runs to the end of the paid period. See the Terms of Service.
Everything in Base with 5× the allowance: 2.5B tokens a month.
Models included
Billed monthly through Stripe, renews until cancelled. Credits never buy the subscription. Cancel any time and access runs to the end of the paid period. See the Terms of Service.
Refer a friend
Invite someone to deeprelay. Once they've deposited $50, $25 of credit lands in your account and $25 in theirs. No cap on how many friends you invite.
Rewards are promotional credit for pay-per-use inference. They can't be withdrawn, don't apply to the subscription, and are reversed if the qualifying deposit is refunded. Program terms
Share your invite link
Every account has one on its Billing page, or from the CLI with deeprelay referral.
Your friend deposits $50
Card, auto-pay or crypto, in one go or over several top-ups.
You each get $25 in credit
Paid the day after they qualify, straight into both balances.
Pay as you go
| Model | Per second | Per clip (from) |
|---|---|---|
| FLUX 3Black Forest Labs · deeprelay/flux-3 | $0.17 | $0.85 |
| Happyhorse 1.0 T2vAlibaba · deeprelay/happyhorse-1.0-t2v | $0.24 | $1.20 |
| Happyhorse 1.1 T2vAlibaba · deeprelay/happyhorse-1.1-t2v | $0.14 | $0.70 |
| Pixverse V5PixVerse · deeprelay/pixverse-v5 | $0.06 | $0.299 |
| Seedance 1.0 LiteByteDance · deeprelay/seedance-1.0-lite | $0.029 | $0.143 |
| Seedance 1.0 ProByteDance · deeprelay/seedance-1.0-pro | $0.113 | $0.565 |
| Seedance 2.0ByteDance · deeprelay/seedance-2.0 | $0.16 | $0.80 |
| Seedance 2.5ByteDance · deeprelay/seedance-2.5 | $0.115 | $0.575 |
| Vidu Q1Vidu · deeprelay/vidu-q1 | $0.044 | $0.22 |
| Vidu Q3Vidu · deeprelay/vidu-q3 | $0.098 | $0.487 |
| Vidu Q3 TurboVidu · deeprelay/vidu-q3-turbo | $0.195 | $0.975 |
| Wan2.7 T2vWan · deeprelay/wan2.7-t2v | $0.02 | $0.10 |
| Model | Context | Input / 1M |
|---|---|---|
| Qwen3 Embedding 8BQwen · deeprelay/qwen3-embedding-8b | 40K | $0.10 |
Prices are USD per 1M tokens for chat and embeddings, and per unit of output for media. The cached input column is the rate for the part of a prompt served from a partner's prompt cache, typically far below the input rate; a dash there means the model publishes no cached rate, so every prompt token bills at the input rate. A dash anywhere is an absent rate, never a zero. An economy row is the same model on cheaper capacity, trading a cold start of roughly 30 to 60 seconds for the lower rate. This page refreshes from the catalog every few minutes.
Questions
Requests keep working. Past the allowance they bill pay-as-you-go at the published per-model rate, the same rates listed below. Nothing stops and nothing queues.
Yes. Upgrade from the Billing page in your console. You pay the bigger plan's full price at that moment and a new monthly period starts right then, with fresh allowances; the unused part of your Base period isn't credited. Every tier covers the same models at the same weights.
No, and the rate tables below say which do. The plan covers the models we serve direct; every other model in the catalog bills pay-as-you-go at its listed rate.
Each model carries a published usage factor. The published factors are printed beside each included model below: a model at ×N draws N tokens of allowance per token served. The factors ship with the plan terms and changes are posted before they apply, so there are no silent multipliers.
Yes, the same as any other input token, and at the model's usage factor. Coding agents resend their whole conversation on every turn, so on a high-factor model a few million tokens of new work can draw a hundred million or more of allowance. Your console meters count these weighted tokens.
Two windows: the monthly allowance, and a cap of 125 million input and 12.5 million output tokens over any rolling 7 days. Usage leaves the 7-day window seven days after it happened, so a heavy day keeps counting for a week. There is no shorter window than that. Past either cap, calls keep working at pay-as-you-go rates while your balance covers them.
Yes, from the Billing page in your console. Your plan stays active to the end of the period you already paid for, and renewal simply doesn't happen.
Yes. Every account has an invite link on its Billing page. Once a friend who signed up through it (or entered your code at sign-up) has deposited $50, you each get $25 in credit. Rewards are promotional credit, spend-only on pay-per-use inference, never apply to the subscription, and are reversed if the qualifying deposit is refunded.
Send us a serious issue report or suggestion from the dashboard and we may award $5 in credit, at our discretion. Not every submission is rewarded, and the program may change or end at any time.
No. The subscription is billed to your card through Stripe. Trial, promotional and referral credits are spendable only on pay-per-use inference.
No. It is the same OpenAI-compatible endpoint either way: same paths, same request bodies, same streaming frames. The plan changes what you are billed, never how you call it.