Sales · why buy voice by the line
You are being billed for how much your customers talk.
Every voice vendor sells the same shape: a meter that charges by the character, so the better your product does, the more it costs you to run. We were that customer. This page is the argument for the other shape.
$0
added when your volume doubles
The bill stops tracking your success
A line is one concurrent call, flat per month. Characters, minutes, voices and seats are not counted at all — so a customer who doubles their talk time doubles your revenue and moves your cost of goods by nothing.
107 ms
to first audio, p50
It answers inside a breath
107 ms to first audio at the median, 108 at the 95th — server time on the production endpoint under call-shaped load, not model-only time. The methodology is published and reruns on demand.
23
languages, one cloned reference
One voice, every market
Ten seconds of reference audio casts a voice that holds its identity across 23 languages. No training job, no per-voice fee, no separate contract to enter a new country.
16
concurrent callers on one GPU
Capacity you can forecast
Lines are provisioned capacity, so the latency figure holds on your first line and your twentieth. Past the edge you get a retryable 503, never a hang — and standby lines fail over on their own.
Take a line and find out.
A key gets you one concurrent line on the production API, with nothing on it counted while you build. No card, no quote gate.
Would rather talk it through first? Book a 30-minute call →
Calendar broken? contact@gandr.ai