TTS for restaurant reservation lines.

A reservation call is ninety seconds of pure turn-taking — date, time, party size, name — where every silence reads as a dropped call. The workload peaks hard at dinner and sleeps at 3 PM, which is exactly the shape a meter prices worst and a line prices flat.

107 ms

first audio — the pause between "table for four?" and the answer

The turn budget

Callers book tables while walking, driving, and half-listening; past a beat of dead air they repeat themselves and bookings double. First audio at 107 ms measured keeps the agent inside conversational rhythm through six or eight rapid turns, in whichever of the 23 languages the caller opens with.

The dinner rush, priced

Reservation traffic concentrates into the two hours a meter counts hardest. On lines, Friday at 7 PM costs the same as Tuesday at 3 — you size to the rush once, and burst lines catch Valentine’s Day at $10 a line-day.

Table 1 — One venue and one group, both models

Operation	Minutes / mo	On a $30–$100 meter	On Gandr lines
One venue, 1 line	3,000	$90 – $300	$150
Five venues, 3 lines	12,000	$360 – $1,200	$450

When every line is mid-call

Past capacity the API refuses fast with a retryable 503 — your orchestrator holds the caller one ring and retries, instead of every active booking degrading at once. The reliability sheet publishes the full posture.

Related sheets.

use case

43,200 min

TTS for AI receptionists

A 24/7 receptionist line speaks up to 43,200 minutes a month: $1,296 to $4,320 on a character meter, $150 flat. The arithmetic for always-on voice.

capability

$10

Burst lines: capacity priced by the day

Traffic past your committed lines spills to burst lines at $10 a line-day, capped in your dashboard — capacity for spike days without paying for them all year.

use case

TTS for hotel front-desk lines

Check-in questions, late checkouts, and wake-up calls in the guest’s own language — a front desk that answers at 2 AM for a flat $150 a line.

Thirty minutes, your production script, the live latency readout — measured in front of you.

Bring this use case to the call