Per-minute usage pricing

A voice agent consumes four metered resources on every call: ASR audio minutes, LLM tokens, TTS characters and telephony minutes. Each has a different unit and vendor. Per-minute pricing rolls them into a single rate per minute of connected call so that a business can forecast cost from call volume and average handling time, and compare directly against the cost of a human agent-minute. Eazyware's voice usage is priced at $0.05 to $0.15 per minute depending on language, model choice and whether components are self-hosted; the platform build is a separate fixed-price project.

The rate is driven by design choices. A faster, smaller model, a cheaper TTS voice, or regional ASR for a single language pushes it towards the lower end; multi-model routing, premium voices and heavy tool use push it up. Silence and hold time count, so a well-designed conversation that resolves in ninety seconds is cheaper than a polite one that takes four minutes.

Per-minute pricing is not the same as the telephony bill from Twilio or Exotel, which is usually passed through separately, nor does it include the Care Plan that covers monitoring and improvement.

• CFO: a single per-minute rate makes the voice channel a variable cost that can be modelled against headcount and volume.
• Founder / CEO: it allows a like-for-like comparison with BPO or in-house cost per call, which is the number the board will ask for.
• CTO: the rate constrains architecture; each vendor swap or model upgrade shows up as a change in cost per minute.
• Support manager: average handling time now has a direct cost line; conversation design that shortens calls reduces the bill.

Buyers cannot reason about tokens, characters and audio seconds across four vendors; they can reason about minutes. Per-minute pricing exists to make the running cost of a voice agent legible and comparable with the human alternative, and to align the vendor's incentives with efficient calls. The trade-off is that the rate is an average: a call with heavy database lookups and a large model costs more than a simple status check, and the vendor bears or passes on that variance. Eazyware publishes a band rather than a single number for that reason and shows the component breakdown when asked.

• Budgeting a hospital appointment line: expected monthly calls multiplied by average minutes and the per-minute rate.
• Comparing an NBFC's outbound reminder cost per contact against a BPO dialler team.
• Setting a usage-based price for a voice feature inside a SaaS product, with a margin over the underlying per-minute cost.
• Modelling a logistics customer-care line's cost during festive peaks when volume triples.
• Choosing between cloud and self-hosted ASR for a bank by comparing per-minute cost at projected volume.

MetricA cost metric you track and forecast, not a skill. Eazyware quotes voice usage at $0.05 to $0.15 per minute on top of the fixed-price build under Voice Agents; see Pricing for the components.

Eazyware service that covers it: Pricing. Starting prices are on the pricing page.