UCaaS pricing bills per user, per seat, per month. CPaaS pricing bills per event: per message, per minute, per API call. Every comparison between the two models reduces to that one difference in the billing unit, and buyers searching "UCaaS vs CPaaS pricing" are almost always asking the same question: which one ends up cheaper for the traffic profile I actually run.
This post does the arithmetic once, in public, using publicly listed figures. The category question is covered separately; the vendor roll-up names the players. Here we only look at the cost model.
The two billing units
Per-seat pricing charges a fixed monthly amount per named user, whether that user handled ten calls or a thousand. The bill is predictable, and it moves only when headcount moves. Every add-on — an AI receptionist, a contact-center module, a meeting assistant — tends to follow the same structure: another fixed amount per seat per month stacked on the base license.
Usage-based pricing charges per event. A call consumes minutes, a text consumes segments, a WhatsApp conversation consumes a session, an AI agent turn consumes model tokens. Headcount is irrelevant; volume sets the bill. The model flips the fixed/variable ratio: cheap when traffic is low or seasonal, expensive only when traffic actually scales, and exactly measurable per channel the whole way through.
The break-even between the two is rarely where the sales material puts it, because the per-seat list price ignores the events that would have been billed under usage-based pricing. Run the full traffic profile across both columns and the gap narrows or flips depending on how often the seats are idle.
A 20-agent model in both columns
Take a support team of 20 agents making outbound calls, receiving inbound calls, sending SMS updates, and running a small set of automated workflows. Approximate monthly usage: 12,000 outbound minutes, 8,000 inbound minutes, 25,000 SMS segments, and 4,000 AI-voice-agent turns. The per-seat column below uses published RingCentral-style list pricing as the well-known category shape (RingEX pricing, roughly $20 to $35 per user per month depending on tier), plus roughly $30 per seat for an AI add-on and roughly $60 per seat for a contact-center add-on on the tiers that require them.
| Line item | Per-seat (RingEX-style) | Usage-based (Orbit example rates) |
|---|---|---|
| Base seats (20) | ~$25 × 20 = $500 | Voice: 20,000 min × $0.014 = $280 |
| AI add-on (20) | ~$30 × 20 = $600 | SMS: 25,000 seg × $0.0075 = $187.50 |
| Contact center (20) | ~$60 × 20 = $1,200 | AI agent turns: usage-billed, ~$40–$120 for this profile |
| Monthly total | ~$2,300 | ~$510–$590 |
The usage-based column prices each channel at the published Orbit rates: $0.014 per voice minute, $0.0075 per SMS segment, $0.01 per RCS message, $0.001 per email, WhatsApp passes through Meta's own fees at zero platform markup, and AI agents bill on usage without a per-seat license. Those are the figures on the pricing page, and they are the only ones used in this comparison. The per-seat column is the arithmetic anyone can reproduce from a RingCentral-style rate card: base license, plus AI add-on, plus contact-center add-on, multiplied by seats.
Even if the per-seat assumption drops to the lowest published tier with no add-ons, 20 seats at $20 a month still land at $400 before a single call is placed. On the usage-based column, those idle seats cost nothing at all — only the actual traffic moves the bill.
When per-seat wins, when usage wins
Per-seat pricing wins when three conditions line up: the team is uniformly busy, every seat makes full use of every add-on, and the monthly bundle price is genuinely a rate the market cannot beat on a per-minute basis. A contact center that runs all 20 agents at full occupancy on every channel every month, and buys the add-on bundles anyway, can come out ahead on per-seat if the license discounts the effective per-minute rate below the prevailing usage-based rate. That shape exists — it is just much rarer than the sales pitch implies.
Usage-based pricing wins under seasonality, under mixed traffic, under automation-first workloads, and whenever headcount is decoupled from traffic. A retail operation that spikes in Q4 and idles in Q1 should not pay 20 idle seats for three months. A product team where AI agents handle 70% of inbound volume should not rent AI add-on licenses for seats the humans never occupy. And any buyer who would rather pay for what they send than pre-pay for capacity they might not use gets a measurably better deal on the usage-based side almost by definition.
The hidden add-on trail
The add-on trail is where per-seat arithmetic quietly breaks. The AI-powered UCaaS post walks the same pattern across the category: RingCentral quotes its AI receptionist (AIR / AIR Pro) and its AI contact center (RingCX) as separate products on top of a RingEX seat, Microsoft stacks Copilot for Teams Phone at roughly $30 per user per month on top of its base Teams Phone license, and Zoom and Dialpad bundle some AI into the plan while reserving deeper coaching features for paid tiers. The trap is the same in each case: the published seat price covers the phone system, and the capabilities a buyer actually wants — AI reception, AI quality scoring, contact-center routing — arrive as additional per-seat line items.
Consolidated, the pattern is simple: per-seat base price + per-seat AI add-on + per-seat contact-center add-on + per-seat Copilot-style add-on = a real monthly bill two to three times the advertised tier price. A usage-based model removes the add-on trail entirely when the AI agents, the contact-center queues, and the voice channels all meter on the same wallet. That is not a claim about which model is morally better. It is a claim about which model a buyer can audit line by line before signing.
Orbit's unified wallet model
Devotel Orbit runs usage-based pricing end to end, and the full category split (UCaaS features and CPaaS channels on one account) is metered off a single prepaid wallet. The pricing page lists one self-serve Pay-as-You-Go track with no monthly fee and no seat count: pre-pay credits, pay only for what you send across every channel, and top up automatically with hard caps if you want that enabled. There is no per-seat AI license on the Orbit side, because the AI agents bill on usage the same way the voice minutes and SMS segments do, and there is no separate contact-center add-on, because call queues and routing capabilities meter on the same wallet as everything else.
That structure is grounded on the live pricing page: Voice $0.014 per minute, SMS $0.0075 per segment, RCS $0.01 per message, Email $0.001 per message ($1.00 per thousand), WhatsApp at zero platform fee over Meta's published rates, and Viber at $0.008 per message. An Enterprise track exists for negotiated volume rate cards, but it is still pay-as-you-go; only the per-message and per-minute rates come down. Nothing on this page invents a subscription tier the pricing page does not publish.
Frequently asked questions
Is UCaaS or CPaaS cheaper?
Neither is cheaper in the abstract. Per-seat UCaaS pricing is predictable when every seat is uniformly busy on every channel. Usage-based CPaaS pricing is cheaper when traffic is seasonal, mixed, or automation-first. The only reliable answer comes from modeling the actual traffic profile on both billing units, which is what this post does for a 20-agent team.
What does a typical 20-agent UCaaS bill look like?
Using RingCentral-style published list pricing, roughly $500 per month for base seats, plus roughly $600 for an AI add-on and roughly $1,200 for a contact-center add-on, for a total around $2,300 per month before usage-sensitive charges. The same traffic on usage-based channels at Orbit's published rates comes in around $510 to $590 per month, depending on AI-agent usage.
When does per-seat pricing actually win?
Per-seat wins when seats are uniformly loaded across every channel every month and the bundle discount brings the effective per-minute rate below the usage-based alternative. That is a real shape, mostly at very high and steady occupancy; it just stops being the default answer once idle seats, seasonal traffic, or AI agents enter the picture.
What is a unified wallet model?
A unified wallet is one prepaid balance that meters every channel — voice, SMS, RCS, email, WhatsApp, video, and AI agents — on the same account with no per-seat license and no per-channel subscription. Devotel Orbit runs its entire pricing page on that structure, so the whole call/SMS/workflow profile pays out of one balance instead of stacking seat-based add-ons.
Does usage-based pricing scale badly at high volume?
High volume still bills at usage-based rates, but volume rate cards lower the per-minute and per-message rates as traffic grows. Orbit's Enterprise track works exactly this way: negotiated rates on the same pay-as-you-go wallet, not a switch back to a per-seat subscription.
The takeaway
Pricing models are not stylistic choices; they are different answers to the question of what the vendor is willing to bill on. Per-seat UCaaS bills on headcount and pre-paid capacity. Usage-based CPaaS bills on the traffic that actually ran. For most teams with mixed voice, SMS, and workflow traffic, the arithmetic favors paying for events, not seats, and the pricing page figures above let a buyer verify that claim instead of taking it on faith.
Published 29 August 2026.