One rate per minute.
One subscription.
One rate for the whole call, because one model handles the whole call, plus one platform subscription sized to your concurrency. Not a transcriber rate, plus a language model rate, plus a synthesiser rate, plus telephony.
From$0.06a minute*
From is the word that matters. $0.06 is the entry rate, not a flat rate. Rates improve with committed volume, agreed per deployment.
One rate per minute
Speech recognition, the language model, the voice, the emotion and the telephony all sit inside the minute. Nothing stacks on top and nothing is billed twice.
One subscription
One platform subscription, from $150 a month depending on the concurrency you need. One platform, one invoice, one team accountable for the call.
Volume discounts
The published rates are the entry point. Rates come down as committed volume goes up, agreed per deployment and written into the contract.
Same platform. Different conversations.
Pick the engine by the work it has to do. Most deployments run both behind one number.
| FLAGSHIPAgni PremiumFrom $0.09 a minute* | Agni V3From $0.06 a minute* | |
|---|---|---|
| Best for | Calls where the conversation carries the relationship or the money. | High volume calling where speed and cost per call decide it. |
| Typical work | Collections and recovery, retention saves, high value sales, VIP and complex service. | Reminders and confirmations, status updates, simple qualification, surveys. |
| Emotional range | Richest range and expression, holds its register through a long, difficult call. | Natural and clear, with more moderated expression. |
| Sentiment reading | Deepest reading, follows how the caller's mood moves during the call. | Core sentiment signals. |
| Dialect coverage | Widest and richest regional coverage we offer. | Core regional coverage. |
| Languages | 50+ languages and regional dialects | 50+ languages and regional dialects |
| Mid-sentence language switching | ✓ Included | ✓ Included |
| Agent memory across conversations | ✓ Included | ✓ Included |
| API and webhooks | ✓ Included | ✓ Included |
| Speech to speech, under 500ms** | ✓ Included | ✓ Included |
| Compliance posture | The same on both. See the trust page | The same on both. See the trust page |
The split between the two is decided in the demo, on your own call mix.
Both engines run on one platform subscription, from $150 a month depending on concurrency. A one-off setup may apply depending on integration complexity, scoped in writing before you commit.
Security, privacy and compliance →
**Response times are from our own testing. Real world latency depends on your carrier and network, the region, and how complex the prompt and tool calls are.
Voice, LLM and emotion. One line on one invoice.
The assembled-stack alternative bills you three or four times for the same minute. We bill you once, and one team is accountable for the whole call. Already on another stack? See how we compare.
What people ask.
By the conversation, not by the feature list. The languages, the agent memory, the integrations, the compliance posture and the response under 500 milliseconds are the same on both. What differs is emotional range, sentiment reading and dialect depth. Routine volume such as reminders, confirmations and status updates runs beautifully on V3. Collections, retention saves, high value sales and VIP service belong on Premium. Most deployments run both, and the demo decides the split on your own call mix.
No. Every inclusion listed on this page is on both engines, including agent memory, API and webhooks, and the same compliance posture. V3 carries more moderated emotional expression and core dialect coverage. If a call type needs the fuller register, route that call type to Premium and keep the rest on V3.
One rate for the whole call. No separate speech recognition, language model, voice and telephony line items that stack per minute, and one platform subscription, from $150 a month depending on concurrency, instead of an orchestration fee that grows with usage. When a call drops, one team is accountable, not four status pages.
Yes. The published rates are the entry point, not the ceiling you are stuck with. Rates improve with committed volume, agreed per deployment against your actual numbers and written into the contract.
It depends on complexity. A straightforward deployment may need little setup. Custom work such as conversation design, dialect tuning on your own recordings, telephony and CRM integration, and a shadow period is scoped in writing before you commit.
Terms are in legal review and will publish here when confirmed. We speak to a 99%+ uptime SLA, agreed per deployment, and we do not claim compliance badges we cannot evidence.
Do your customers feel it?
Bring your ops lead. We call your own front desk in your market's dialect, live on the demo, and size the rate to your volume while you watch.
Want a number before you talk to us? Run the ROI calculator
*Indicative prices, not a proposal or offer. Pricing varies by region, market and solution requirements, and final terms are agreed in contract. A platform subscription from $150 a month applies, depending on concurrency, and a one-off setup may apply depending on integration complexity. Calculator outputs are estimates built from the inputs you supply and are not a quote.