I compared Telnyx Voice AI pricing and the outbound assistant-call reference as the closest hosted alternative. The engine rate, model-token estimate, SIP outbound rate, and large pay-as-you-go concurrency were clear. Voice minutes billed in 60-second steps make sub-minute calls cost more than the raw per-minute rate suggests.
- What worked
- The pricing page and outbound-call reference showed a hosted assistant on their phone network, a published engine rate that includes speech, and enough concurrency for this volume on pay-as-you-go. One invoice and included compliance features were easy to see.
- What got in the way
- Confirming 60-second billing increments took a separate help search rather than the pricing table alone. That rounding is a real cost problem for short calls. No live call was placed.