Skip to content

The price is the price

No seats. No platform fee. No token math. Telephony is listed separately on purpose.

Omni

$0.05per minute, per-second metering
$3.00 per hour for speech + brain
  • One speech + brain rate: hearing, reasoning, retrieval, tool calls, and speech
  • Native Omni realtime API and managed WebSocket
  • Managed telephony is a separate $0.01/min line item

Use a sandbox key for capped testing without billing. Fund a live key before production traffic.

A 4-minute Omni session costs $0.2. Managed telephony adds $0.04.

Every number above is the exact rate we bill from, the same figure on your invoice.

The terms, once

  • $50.00 free credit at signup, no card; a sandbox key never bills
  • Per-second metering, summed across the month and rounded once at the invoice; telephony is the only 1-minute pulse
  • No seats, no platform fee, no per-call minimums, no annual lock-in
  • No egress charges, no overage tiers, no support paywall
  • OpenAI-compatible: point your existing SDK at api.pyai.com
  • Commit volume and the rate goes down, never up
You’ve spent reading this
$0.0000

Billed by the second at Omni’s $0.05/min. A 1-minute-minimum vendor would already have charged you $0.05.

AI usage has no per-call minimums.

Most voice platforms round every call up to the next minute, a habit inherited from telephony. On a fleet of short calls, that quietly multiplies your bill. For AI usage, PyAI sums exact seconds across the whole billing period and rounds once, at the invoice line.

Typical: rounded per call
$5,000

100,000 calls × 12s, each rounded up to a 1-minute minimum = 100,000 billed minutes.

PyAI: summed, rounded once
$1,000

100,000 calls × 12s = 20,000 actual minutes, billed as metered. No per-call rounding.

Same calls, same rate, 5× the bill comes from rounding alone. That gap is invisible until you read the invoice.

Illustrative at $0.05/min for Omni speech + brain. Managed telephony is separate at $0.01/min and keeps a 1-minute pulse because resold carrier minutes have different economics.

The card is the ceiling

You don't pick a plan. Start free on the printed rates, commit at volume to go below them, and bring in Enterprise when procurement calls.

01 · day one

Developer

Free to start

$50.00 credit, no card, the printed rates.

10 concurrent calls · 20 req/s · 10 cloned voices

02 · at volume

Business

$1,000/mo committed

Rates drop below the card. Priority support.

50 concurrent calls · 50 req/s · 100 cloned voices

03 · at procurement

Enterprise

Custom

SSO, BAA, SLA, dedicated review workflows.

1,000 concurrent calls · 1,000 req/s · unlimited voices

Pricing FAQ

How does PyAI pricing compare with Vapi?

PyAI Omni is $0.05/min for speech + brain plus a separately listed telephony rate. Vapi advertises about $0.05/min as a platform fee; all-in typically lands around $0.10-$0.31/min once you add STT, an LLM, TTS, and telephony.

Are there per-call minimums?

No. Most voice platforms round every call up to the next minute, which inflates the bill on short calls. PyAI sums your exact billable seconds for each product across the whole billing period and rounds once, at the invoice line - never per call. A fleet of short calls is billed on its summed minutes, not minute-rounded individually.

What's in the free tier?

$50.00 in free credit at signup, no card. Spend draws down the credit; when it runs out you simply add funding. AMD's first 5,000 answered calls each month are free.

Can I go live without a card?

Your sandbox test key works instantly and free. When you're ready for live traffic, add a card or top up prepaid credit in the console - self-serve, no sales call. Phone verification unlocks free live credit first.

Are there volume discounts?

Yes. The rate card is the ceiling, not the floor: Business and Enterprise commit volume and pay less. Batch transcription is already half price at $0.0005/min, and Trace can be packaged as an annual compliance plan.

Get an API key. Make a phone agent.

Sandbox keys never bill. Fund a live key when you go to production.