Comparison
Per-minute or flat-rate — how should an AI voice agent be priced?
Most of the AI voice market bills per minute, at roughly $0.07 to $0.31 all-in once speech recognition, the language model and text-to-speech are stacked on the platform fee. Flat-rate plans do exist, but in most cases the meter reappears somewhere else — on concurrency, on model spend billed to your own provider account, or as a plan tier sized around a fixed number of answered calls. Per-minute suits low, predictable volume; flat-rate only saves you money if it stays flat at the volume you actually intend to run.
Note · Competitor rates below were verified against published pricing pages in August 2026 and change frequently. Treat this as dated evidence, not a live comparison, and check current pricing with any vendor before deciding.
Last reviewed
The four billing models
Almost every vendor uses one of four shapes. Knowing which one you are being sold matters more than the headline number, because they fail in different places as you scale.
| Model | How it works | Suits | Where it bites |
|---|---|---|---|
| Per-minute | A platform rate, usually with speech, model and voice costs added on top | Low or unpredictable volume; pilots | Cost scales linearly forever, and the advertised rate is rarely the all-in rate |
| Minute bundle + overage | A monthly fee including N minutes, then a per-minute rate beyond it | Steady, well-understood volume | The overage rate is where the margin sits — check it before the bundle size |
| Per-channel | A flat fee per simultaneous call, with unlimited minutes on each | Predictable concurrency, like a small reception line | The meter moved to concurrency. A surge needs channels you are paying for year-round |
| Flat subscription | A flat fee with calling not billed per minute; throughput sized by concurrency | Outbound at scale, where per-minute costs would dominate | Only genuinely flat if model and infrastructure costs are included, not passed through |
What the market actually charged in August 2026
The platforms below are alternatives to CallableAI, not parts of it — CallableAI runs on its own telephony and orchestration and does not resell any of them. From a scan of published vendor pricing pages; independent teardowns put the real all-in cost of per-minute platforms between roughly $0.07 and $0.31 per minute once every component is counted.
| Platform | Model | Published rate | Unmetered? |
|---|---|---|---|
| Vapi | Platform fee + model passthrough | $0.05/min + speech/model/voice on top (~$0.15–0.31 all-in) | No |
| Bland | Bundled per-minute | $0.11–$0.14/min | No |
| Retell | Itemised per-minute | ~$0.07–$0.31/min | No |
| Vociply | Minute bundle + overage | $299–$749/mo, then $0.25–0.26/min | No |
| Nedzo | Outcome-based | $299–$499/mo, then $1.99/outcome | No — metered on outcomes |
| QCall.ai | Per-channel unlimited | $549 per channel/mo, 1 concurrent call | Partly — capped by paid channels |
| OpenVoice Agents | Flat infra, bring-your-own-key | $199/mo unlimited infra minutes | Partly — model spend billed to you |
| KaiCalls | Flat monthly receptionist | From $69/mo, tiered by call count | Partly — plan tiers by volume |
| CallableAI | Flat subscription, throughput by concurrency | Flat monthly fee; calling not billed per minute | Yes |
To be straight about our own position: flat-rate voice AI is not unique to CallableAI, and we do not claim it is. What separates these offers is where the meter reappears. QCall.ai meters concurrency at $549 per simultaneous call. OpenVoice passes model spend through to your own provider account. KaiCalls sizes plans around roughly 150 answered calls a month for a single business line. None of those is an unmetered outbound engine — which is the specific thing that matters if you intend to run volume.
Do the arithmetic on your own volume
The crossover is easy to compute and worth computing before any sales call. Take your intended monthly connected minutes, multiply by the all-in per-minute rate — not the advertised platform rate — and compare against the flat plan.
- At 500 minutes a month, per-minute at $0.15 all-in is about $75. Almost any flat plan loses.
- At 5,000 minutes, the same rate is about $750. The comparison gets interesting.
- At 50,000 minutes, it is about $7,500 a month, and a flat plan wins decisively — provided it is still flat at that volume.
The trap in that last line is the qualifier. Ask any flat-rate vendor what happens at ten times your current volume, and whether the answer involves buying more channels, more plan tiers, or a bigger model bill in your own provider account.
The number that settles it
Neither model tells you what a customer costs. Cost per qualified lead does, and it is driven far more by list quality and offer than by billing structure. Across three CallableAI tenants in one week, a qualified lead cost $18.89 blended — but individual tenants ranged from $17.11 to $44.43 on the same platform, in the same week.
A 2.6× spread on identical pricing is the strongest argument we know against choosing a vendor on their rate card. Work out your cost per outcome on a trial, at your own volume, with your own list.
Common questions
- What does an AI voice agent cost per minute?
- Roughly $0.07 to $0.31 all-in across the per-minute vendors surveyed in August 2026. The advertised platform rate is usually much lower than the all-in figure, because speech recognition, the language model and text-to-speech are billed separately on several platforms. Ask for the all-in rate on a representative call, not the headline.
- Is unlimited AI calling really unlimited?
- Usually not in the sense you would assume. Plans advertised as unlimited commonly meter somewhere else — per simultaneous channel, or by passing model costs through to your own provider account, or by tiering plans around an expected number of answered calls. The question to ask is not whether minutes are unlimited but what happens at ten times your current volume.
- Is Callable the only unmetered AI voice platform?
- No, and we correct that claim when we see it. Several vendors sell flat-rate or per-channel plans. What we have not found elsewhere is calling that stays unmetered at outbound scale — no per-minute line across hundreds of thousands of dials a week — which is a narrower claim than being the only flat-rate option.
- Which pricing model is cheapest?
- Entirely a function of volume. Per-minute is cheaper at low volume and worse at high volume; the crossover for most businesses sits in the low thousands of minutes a month. But billing model is a second-order question — cost per qualified lead varies more between two businesses on the same plan than between two pricing models.
Deployed white glove. Backed by an SLA.
CallableAI works with organisations that need quality, volume and compliance to hold at once. Bring us your requirements — we'll scope the deployment, the rollout and the SLA with you.