Pricing

Modelworks is a speech API. Text in, audio out, billed per character. No subscription, no minimum, no seat.

The price

No charge for the first 250,000 characters - four to six hours of audio on the conversion table below, enough to know whether the quality is good enough.

Buy one million characters for $12 - prepaid, no subscription, and the key arrives by email within a minute or two.

The conversion, stated so every figure above can be checked. How many characters make an hour of speech depends on how fast the voice reads, and the industry does not use one number:

Taking 20 hours as the midpoint: 50,000 characters an hour, so $0.60 an hour and one cent a minute. At the ends of the range it is $0.71 an hour if your voice reads fast and $0.52 if it reads slowly. A customer generating an hour of audio a day pays about $18 a month - between $15.65 and $21.18 across that range.

Where $12 sits

Every figure below is from the vendor's own pricing page, read 6 to 9 October 2026.

Where that puts us: above the commodity tiers, below the managed ones. The table is the market as published, so you can place us against whatever you use today. What $12 buys is not a lower per-character rate - it is the four reasons below, and they are the difference between a commodity endpoint and this one.

$12 is for the buyer whose problem is not the price. It is parity with Everlit on Indic and a quarter below Polly Neural and Google's mid tiers, and the four reasons below are what the difference buys.

The four reasons

1. Your content is not retained. Text you send and audio you get back are discarded when the request completes. Request metadata only, 30 days. The cheapest endpoints keep content and several say so in their own terms; if your content is regulated, customer-owned, or simply not yours to leave lying around, that difference is the entire purchase.

2. A published DPA that incorporates the EU Standard Contractual Clauses, Module Two. It is on the site, with its security annex and transfer assessment, and it takes effect with the terms - no sales call and no separate signature. India is not an EU adequacy country, so an EU buyer cannot use a cheap endpoint without one at all.

3. A sub-processor list short enough to read in a minute. No analytics, no error-tracking SaaS, no support platform, no LLM provider. Every row says what that party can actually see.

4. Somebody answers. A named operator who replies to email and ships the fix.

Those four are the product. Everything else about a text-to-speech endpoint is commodity, and priced like it.

What we do not claim

We do not offer an SLA. We will not promise an uptime from a small number of hosts that we cannot control. A status page with real incident history is the honest substitute, and it is linked from this site.

Tuned for throughput and cost per hour, not for milliseconds. The production box measures 2.85x realtime at four concurrent requests, and the status page carries the current figure rather than a promise. This API is built for latency-tolerant workloads - audiobooks, e-learning, podcast repurposing, archives - not for sub-100ms voice agents.

We do not have SOC 2. It is commercially triggered rather than legally required at our size. We will get it when a contract depends on it and say so here when we do.

Volume

Below 25 million characters a month, pay per character. At or above it, a committed block is the better deal - a flat monthly rate for a floor you can rely on, priced under the per-character rate. Ask and we will tell you which side costs you less.

There is no second threshold and no band in between. A committed block is a discount you opt into, not a floor you are held to: stop using it and you revert to per-character pricing.

Billing

Pay by card. Invoices and receipts come from our payment provider, which handles sales tax, VAT and GST for you rather than by you.