Guide · AI Receptionists
When two callers ring at once: what concurrency actually costs as you scale a voice-AI platform
The headline per-minute rate ignores the line that bills you for simultaneous calls — Vapi and Bland price it in opposite ways, only Vapi publishes a price past 50 lines, and Synthflow no longer publishes one at all.
The per-minute rate everyone quotes assumes one caller at a time. Real phone traffic doesn’t arrive politely in single file — it clusters at lunch, after an ad runs, on Monday morning — and the moment two callers ring simultaneously you need a second concurrent line. Almost no buyer models this, because most vendors don’t print it on the pricing page. Three developer-grade platforms in our dataset used to, each pricing the extra line a different way. Now only two still publish a figure, and one of those stops at 50 lines.
The direct answer
If a second caller rings while the first is still talking, here is who charges you and how:
| Platform | Concurrency included | Cost of the next line | Per-minute talk rate |
|---|---|---|---|
| Vapi | 10 lines | +$10/line/mo | $0.05/min (platform fee) |
| Bland | 10 / 50 by tier | bundled into tier ($0 / $299); above 50, Enterprise, no published price | $0.14 / $0.12/min |
| Synthflow | not published | not published — Enterprise contracts from $30,000 annually | not published |
Vapi bundles 10 concurrent lines into its Build (pay-as-you-go) plan and charges $10 per additional line per month above that. Bland takes the opposite approach: it bakes concurrency into the tier itself — 10 concurrent calls on Start (free) and 50 on Build ($299/mo). Its Scale tier, 100 concurrent calls for $499/mo, was last seen on 2026-09-02 and is no longer listed as of 2026-09-27; the page’s plan comparison now reads Start, Build, Enterprise. Synthflow used to sell concurrency as a metered add-on at $20 per unit per month on its $0.09/min pay-as-you-go platform, but that plan, last seen 2026-06-27, is no longer listed either.
Why the cheapest headline can carry the most expensive scaling penalty
For a solo practice answering one or two calls at a time, both published options are effectively free on concurrency — Vapi’s 10 included lines and Bland’s free-tier 10 cover you. The divergence only appears once you grow.
Walk a busy inbound operation up to 50 simultaneous lines. On Vapi that is 40 lines above the included 10, at $10 each — $400/mo in concurrency fees alone, before a single minute of talk time. Bland’s Build tier hands you exactly 50 concurrent calls for a $299/mo platform fee, and its all-in talk rate ($0.12/min) already covers model, speech, and telephony. The platform that looked cheaper per minute ($0.05 vs $0.12) is now $101/mo more expensive on the line fee before usage even enters the math.
Push to 100 concurrent lines and only one of the three still prints a price. Vapi: 90 extra lines × $10 = $900/mo. Bland used to answer with Scale at $499/mo flat, roughly half that — but with Scale off the page, anything past Build’s 50 lines is an Enterprise conversation with no published figure. Bland’s bundled curve is the flatter one for as far as it is published; past 50 lines, Vapi’s meter is the only number you can put in a spreadsheet.
Synthflow has left the published comparison altogether. Its pricing page now sells only Enterprise contracts that “start at $30,000 annually,” with final pricing “scoped around call volume, concurrency, telephony setup, integrations, security needs, and launch support.” That floor alone is $2,500/mo ($30,000 ÷ 12) before you know what a line costs, and concurrency has become one of the terms you negotiate rather than a unit you provision.
The number the per-minute table hides
The lesson isn’t that one model wins. It’s that concurrency and per-minute rate move in opposite directions between Vapi and Bland. Vapi has the lowest platform fee ($0.05/min) and the cheapest single extra line ($10), yet its metered model compounds into the steeper bill at scale because every line is additive forever. Bland charges the higher per-minute rate ($0.14/min on its free tier) but the cheaper path to 50 concurrent calls, because the lines come free inside a fixed tier — and beyond 50 it stops publishing a price.
A buyer who picks on the per-minute sticker alone — Vapi at $0.05 — and then scales to a real call center load can quietly add $400–$900/mo in line fees that never appeared in the quote. A buyer who reads concurrency first sees that Bland’s higher minute rate buys a flatter, more predictable curve up to 50 lines, and a sales call after that.
How to read this
Model the peak, not the average. If your traffic rarely exceeds a handful of simultaneous calls, Vapi’s 10 included lines and lowest per-minute fee make it the cheapest answer, and Bland’s free tier matches the line count at $0. If you expect sustained bursts into the dozens of concurrent calls, Bland’s Build tier (50 lines at $299) is the flatter curve, and Vapi’s $10/line meter is the one that punishes growth. Past 50 lines, Vapi is the only one of the three that still publishes what the next line costs; Bland’s Enterprise tier and Synthflow’s Enterprise contracts (from $30,000 a year) both put concurrency behind a quote. Whichever you pick, the concurrency line — not the per-minute rate — is what sets your bill once the phones get busy.