Pricing

Pay only for what you use.

No monthly fee. No minimums. No commitment. Drop in a card, run one call, pay for one call. Cancel any time. The free sandbox tier never expires.

Text models

Per-token pricing, every tier priced separately.

Each text tier has its own input and output rate. Pick the tier that matches the job; you're billed for the tokens that move through the API.

Model Input Output Best for
qai-hello-world Free Free Free sandbox tier (100/key/day cap)
qai-flash $0.20 $0.50 Cheap and fast (per 1M tokens)
qai-pro $0.80 $2.00 Balanced default (per 1M tokens)
qai-max $1.50 $5.00 Top capability (per 1M tokens)
qai-think $1.80 $8.00 Reasoning (per 1M tokens)

What does this cost in real money?

Imagine you are running a chatbot at 100,000 calls per month, with an average of 400 input tokens and 200 output tokens per call. That is 40M input tokens and 20M output tokens monthly. Here is the bill on each text tier:

qai-flash
$18.00
/ month
qai-pro
$72.00
/ month
qai-max
$160.00
/ month
Image models

Per-image pricing.

Both image tiers bill per image returned. If you set n: 4, you get four images and you are billed for four. Permanent CDN hosting included when you pass hostMedia: true.

Model Price Best for
qai-imagine-turbo $0.04 Real-time, batch, preview workflows
qai-imagine-quality $0.08 Studio-grade detail and photorealism

1,000 images per month

If you generate a thousand images monthly, here is what that runs:

qai-imagine-turbo
$40.00
/ month
qai-imagine-quality
$80.00
/ month
Video models

Per video-second.

Billed on the duration returned, not the duration requested. A 6-second clip bills 6 seconds; a 4-second clip bills 4. Hosted on the Qai CDN out of the box.

Model Price Best for
qai-motion $0.18 Up to 6-second clips, 832x480

100 video-seconds per month

Roughly 16 six-second clips, or 33 three-second clips.

qai-motion
$18.00
/ month
Utility endpoints

Free with any Qai key.

Stateless helpers like Clean JSON and Humanize Text are free for every account. They do not consume LLM tokens. Rate-limited per key on the same account-wide allowance as your other endpoints.

Clean JSON

POST /v1/utilities/clean-json — Free. Strips markdown fences, fixes trailing commas, converts Python booleans, recovers valid JSON from messy LLM output.

Humanize Text

POST /v1/utilities/humanize-text — Free in default mode (typography cleanup). Optional naturalizeWording mode uses a small text model and is billed at qai-flash rates.

Spend controls

You decide the ceiling.

Three layers of cost control to make sure you never get a surprise invoice.

Per-API-key budgets

Each API key can have a daily and a monthly spend cap configured from the dashboard. When the cap is hit, that key starts returning 402 Payment Required. Your other keys keep working. You get an email at 50%, 80%, and 100% of the cap.

Live usage dashboard

Your dashboard shows running spend in real time, broken down by model, by API key, and by time range. Spot a runaway cron job an hour after it starts, not at the end of the billing cycle.

The free sandbox tier never expires

qai-hello-world is unlimited in development and capped at 100 calls per key per UTC day in production. Use it during integration work so your real spend only kicks in when you go to production.

Billing FAQ

Questions about how the meter ticks.

How am I billed?
Stripe charges your card at the end of each monthly billing cycle for whatever you used during the cycle. No prepayment, no monthly base fee. The invoice is emailed automatically and available in your dashboard.
What counts as an input vs output token?
Input tokens are everything you send to the model: system prompts, user messages, prior conversation history, any embedded context. Output tokens are what the model generates back. Both are counted by the upstream tokenizer that the model uses.
Are there minimum spend requirements?
No. The minimum is $0. If you run one API call this month, you get a $0.0001 invoice.
Do you charge for failed requests?
No. A 4xx or 5xx response doesn't bill. Cancelling a stream mid-flight only bills the tokens that already came through.
Do you offer volume discounts?
Not at the SKU level for v1. As you scale, contact us and we can usually structure something. The pricing on this page is the published rate for everyone.
What about taxes?
Stripe handles tax collection. Canadian customers see GST/HST/PST applied based on province. International customers see VAT or local sales tax where applicable.
Can I get an invoice for accounting?
Yes, automatically. Every cycle Stripe generates an invoice and emails it to the address on your account. Past invoices are downloadable from your dashboard.
Can I switch tiers mid-cycle?
There are no "tiers" in the subscription sense - you pay per call. Switch models in your code whenever you want; you are billed at that model's published rate from that call forward.
How does Qai's pricing compare to other AI APIs?
Across the board, Qai sits 25-50% below comparable models from the big AI labs. The trade-off is that we are a younger company - if you need a SOC 2 report, enterprise SSO, or a 99.99% SLA today, our larger competitors are a safer bet. If you want the same quality for less money and you can live with a smaller vendor, that is where Qai wins.
What happens if my card fails?
We retry it three times over the next week. If it still fails, your API keys get a "past_due" status and start returning a billing-error response. Update your card from the dashboard to reactivate.

Predictable pricing.

Sign up free. The sandbox tier lets you test your integration without spending a cent.

Get your API key