Synthflow Pricing 2026 for 1K, 10K & 100K Minutes
Nov 11, 202615 min read
Synthflow sells a short ladder of plans: self-serve tiers with bundled minutes, sales-quoted contracts, and overage and concurrency limits sitting outside the headline price, which is the part teams wiring calls into a workflow built on Activepieces tend to notice only after the first invoice.
Synthflow is a no-code platform for AI voice agents. Whoever gets the page when a call queue stalls at 2am will find that limit in the plan, not in the invoice total.
Synthflow pricing today: every plan in one table
The plan table, checked 2026-11-11
Read the minutes, overage and concurrency columns first, because those decide what a month costs. The table lists the plans currently sold on synthflow.ai and how each one handles the three limits that move the bill.
Retired self-serve plans are left out.
| Plan | Price | Minutes | Overage | Concurrency | Key features |
|---|---|---|---|---|---|
| Starter | Self-serve, billed by card | Bundled allowance | Billed per minute past the allowance | Low, suited to one agent | Entry tier for building and testing |
| Pro | Self-serve, higher monthly fee | Larger bundled allowance | Billed per minute past the allowance | Higher than Starter | Fuller build tools for live deployments |
| Enterprise | Quote-only, annual contract | Negotiated | Negotiated | Negotiated | Contract terms, support and security terms agreed with sales |
When a row is quote-only, the number comes out of a sales call, so someone on your side must own that negotiation and the renewal date.
You will hit the entry tier's limits on minutes and simultaneous calls first, because it exists to build and test an agent, not to carry production traffic. So treat it as a prototype environment and price the live deployment from the paid rows.
Sign annually and you exchange a lower effective rate for a fixed commitment. Enterprise is sold that way by default.
If you are running this arithmetic for your own team, see what the same workload costs on Activepieces.
Synthflow pricing at volume tiers
Whoever owns the allowance owns a bill with two parts. The plan fee is set by Synthflow, while telephony plus model usage scales with every minute and no plan size removes it.
$0.04 to $0.07 per minute is the second part, before the plan fee is counted. Telephony is $0.02 per minute and the language model $0.02 to $0.05, so the AI component costs as much as or more than telephony per minute.
At each volume, the table shows that pass-through cost. Read it as a floor: your all-in minute can never come in below it, and the plan fee and any overage sit on top.
| Monthly minutes | Telephony ($0.02/min) | LLM ($0.02–$0.05/min) | Pass-through per month | Floor per minute |
|---|---|---|---|---|
| 1,000 | $20 | $20–$50 | $40–$70 | $0.04–$0.07 |
| 10,000 | $200 | $200–$500 | $400–$700 | $0.04–$0.07 |
| 100,000 | $2,000 | $2,000–$5,000 | $4,000–$7,000 | $0.04–$0.07 |
Look at the floor per minute: it is identical in every row, so volume doesn't make the pass-through cheaper. Only the plan fee and overage can be spread thinner.
To find your all-in minute, add the plan fee and any overage to the pass-through figure, then divide by minutes used.
At 1,000 minutes a month the pass-through is $40 to $70. A plan fee larger than that means the plan, not the calls, is most of what you pay.
At 10,000 minutes the pass-through reaches $400 to $700, and usage, not the plan fee, drives the monthly bill, so controlling your bill now means controlling how much you use.
The risk moves to the allowance: if usage crosses it, Synthflow bills overage per minute on top of the floor, and the person watching usage is the one who sees it first.
At 100,000 minutes Synthflow quotes the volume as custom, so there's no published plan fee to calculate with. The $4,000 to $7,000 pass-through is the only part you can price in advance, which makes the quote the figure to negotiate.
Hidden costs and usage limits
Before you read the plan price as the bill, know that it leaves out at least five line items, and sales quotes two of them, so the quote is where you find them.
At 2am, when an outbound campaign meets the cap and calls queue or drop, concurrency is what fails. It is the number of calls running at once. Someone has to own that alert.
Extra slots cost $20 per slot per month, so ten more lines add $200 a month before a single minute runs.
Telephony, numbers and third-party model charges
$0.02 per minute is what managed Twilio telephony costs. Twilio lists US inbound at $0.0085 and outbound at $0.014, so the managed fee is roughly 2.4 times the carrier's inbound rate and about 1.4 times its outbound rate.
Bringing your own carrier changes the figure. Twilio reports that Telnyx lists $0.0032 inbound and $0.005 outbound, so at 10,000 outbound minutes that is $200 managed against $50 on Telnyx, a $150 monthly difference.
The language model adds $0.02 to $0.05 per minute, or $200 to $500 at the same volume.
Synthflow voices, languages and white-label pricing
Check voice and language counts against the plan you are buying, because they aren't uniform across tiers.
White-label, which removes Synthflow's branding for agencies reselling the product, runs about $2,000 a month, or $24,000 a year, so resellers must factor this into their customer pricing.
If you run a healthcare team on a large plan, this next one will hit hardest: a HIPAA business associate agreement carries a surcharge of about 30%, per Agentsbyaiq.
| Line item | Cost | Status |
|---|---|---|
| Extra concurrency | $20 per slot per month | confirmed |
| Managed Twilio | $0.02 per minute | confirmed |
| LLM | $0.02 to $0.05 per minute | confirmed |
| HIPAA BAA | about 30% surcharge | ask sales |
| White-label | about $2,000 per month | ask sales |
Treat the two "ask sales" rows as negotiable ranges, not rates.
I couldn't confirm an overage rate or a minimum term, so both belong in writing before you sign. Without them, a month that runs past the included minutes is billed at a rate you haven't seen.
The business case for Synthflow
If you pay for Synthflow, the best argument is the engineering time it saves. A month billed at an unfamiliar overage rate is still cheaper than the engineering weeks you would spend avoiding it, provided those weeks are scarce.
Why Synthflow's higher price can be worth it
Start with who gets paged. In a hand-built voice agent, the person who wrote the telephony glue, the speech handling and the retry logic is the person woken when a call drops at 2am on a Tuesday.
If that person is a contractor who has since left, nobody owns the failure.
With a no-code builder, part of that ownership moves to the vendor, and a non-engineer can change a script without a deployment.
That's the real saving, and it's larger than the build time alone. A working agent in an afternoon means a team can test whether callers will talk to it at all before committing a developer to it.
If the answer is no, the platform fee was the price of finding out cheaply.
That's the case for a business with no spare engineers, or one that needs a pilot live this month.
Where Synthflow pricing stops paying off at volume
It only holds while the premium stays small next to the labor it replaces. That stops being true as volume rises.
You pay engineering cost once; the all-in minute, with overage, telephony and add-ons counted, is paid on every call, indefinitely. At high volume the gap between that minute and an API-first rival's compounds each month, and a developer's build time is repaid quickly.
You pay engineering cost once; the all-in minute, with overage, telephony and add-ons counted, is paid on every call, indefinitely.
When a developer is already on staff, the no-code advantage also shrinks. Their time to a first working agent is longer, but it is a one-off, and they leave you with code your team owns and can move between vendors.
The glue around the call is the one part you can buy cheaply either way.
Activepieces can start the Synthflow call from a CRM event and write the result back to the record, and the run is the meter, not the steps inside it: a flow costs 1 credit whether it has two steps or ten, which means your costs don't scale with workflow complexity, where Zapier bills per task and Make per module.
Splitting the work into smaller, safer steps therefore never raises the bill. MoneyGram, Moneypenny, Alan and FundingSocieties run it in production. That prices only the plumbing; the all-in minute stays Synthflow's.
It comes down to one condition: whether you have a developer available and the call volume to justify using them. If you have neither, pay for speed. If you have both, build, and price the minute before you sign.
Worth checking against a plan that does not meter every step: one credit covers a whole run on Activepieces.
Synthflow versus Retell, Vapi and Bland on price
Compare the headline rates only after you know what each one leaves out: the four platforms don't quote the same minute. Synthflow bundles it into a plan, Vapi quotes a platform slice, and Retell and Bland sit between.
Read the table below for which column your invoice will move on, not which row looks cheapest. It sets the four vendors' framing side by side, taken from the pricing logic each publishes.
| Platform | What "per minute" means | Billed separately | Platform fee | Free credits | Concurrency | Build style |
|---|---|---|---|---|---|---|
| Synthflow | Minutes drawn from a plan allowance, overage after | Telephony and add-ons | Yes, tiered plans | Trial offered | Capped by plan | No-code |
| Retell | A range that moves with model and voice choice | Telephony in some setups | Usage-led | Trial credit | Included lines, more on request | API-first, with a builder |
| Vapi | An orchestration rate only | LLM, voice, telephony at provider cost | Usage-led | Trial credit | Default cap, raised on request | API-first |
| Bland | A per-minute rate tied to plan tier | Varies by tier | Yes, tiered | Free tier | Tiered by plan | API-first, with a flow editor |
Confirm every cell against the vendor's own page on the day you sign.
When a rate omits the language model, the voice and the phone line, it is a component price, not a call price. Vapi's is the clearest example. Add your own sources for the missing parts before comparing it with a bundled rate.
Don't read free credits as more than proof that the demo works.
Concurrency is what pages someone at 2am when a campaign queues calls behind a cap, so find out who owns raising it and how quickly the vendor will do so. Plan fees decide whether idle months cost you anything.
If you are not an engineer, Synthflow's builder buys you speed, and you pay for it in the all-in minute. The API-first rivals ask for engineering time up front and reward it with component-level control over cost.
Synthflow calls from workflow events
You still need a little engineering, but far less: a CRM or form event sends one API request to start a call, and one webhook brings the result back. Whoever owns that pair gets paged when it breaks.
Starting calls and receiving results through the API
When a lead is created or a form is submitted, your workflow sends a request to Synthflow's documented API. The request names the assistant and the number to call, and passes the context the agent needs.
The result path runs the other way. Point Synthflow's webhook at an endpoint you control, and write the outcome back to the CRM record: transcript, status, any captured fields.
Zapier, Make and n8n all handle this with an HTTP request step going out and a webhook trigger coming in. A direct call from your own backend does the same job with more control over retries.
Decide who owns the failure cases first. A call that never connected, a webhook that arrives twice, and a result that can't find its CRM record are what turn up at 2am. Each needs a named owner and a log you can read.
The same two moves apply in any visual builder that offers an HTTP step and a webhook trigger. Build the request and the receiver once, then test both before adding branching logic.
The order matters, because each step makes the next one cheaper to do properly:
- Reread the pricing page. Plans change, and the all-in minute you priced last quarter may not be the current one.
- Price your real minutes. Use your own call lengths, retries and peak concurrency, not the plan's headline allowance, and include overage, telephony and add-ons.
- Test on the entry plan. Run the CRM trigger and webhook path end to end, so the first bill you see comes from real calls.
- Set spend alerts. A looping workflow that redials is an invoice, not a bug report, so cap it before it runs unattended.
If the all-in minute still works after step three, the no-code speed is paying for itself.
What Activepieces does about this
Activepieces does not change Synthflow's minute. The all-in figure still comes from Synthflow's plan, overage, telephony and add-ons, and nothing on our side lowers it.
What we change is the cost of the glue that article describes: the CRM trigger that starts a call and the webhook that writes the result back.
In Activepieces that pair is one flow, built with the HTTP integration going out and a webhook trigger coming in.
A flow run costs 1 credit whether it has two steps or ten, so adding a duplicate-webhook check or a "no matching CRM record" branch doesn't raise the bill.
The flow is also where you put the guardrails from the checklist above.
You can count redial attempts per lead and stop at a limit you choose, or send a Slack or email alert when a run fails, so the 2am failure lands with a named person and a run log rather than in a month-end invoice.
We are not a substitute for Synthflow's own spend alerts, and you should set those too. They cover different failures: theirs watch the minutes, ours watch your workflow.
The licence matters if you are pricing a pilot against an API-first rival. The community edition is MIT-licensed open source, so you can self-host the glue and keep it when you change voice vendors.
If you build on Synthflow today and move to Retell or Vapi next year, you rewire one HTTP step rather than rebuild the automation around it. MoneyGram, Moneypenny, Alan and FundingSocieties run Activepieces in production.
Test the flow on Synthflow's entry plan first, as step three says, so the first bill you read comes from real calls.
Frequently asked questions
After the all-in minute has been calculated, a finance reviewer raises the questions below. The article doesn't settle them with figures, because each answer sits on a Synthflow page that can change.
Free plans and cancellation
Does Synthflow have a free plan? Don't assume one. Whether Synthflow currently offers a free tier or only a time-limited trial is stated on its pricing page, and that page governs.
Whichever it is, ask who owns the test account, because a prototype that lapses mid-pilot becomes the evening's incident.
Can I cancel Synthflow or change plans at any time? Read the billing terms before the first invoice, not after. Plan changes and cancellation rules are set out in Synthflow's terms and on its pricing page.
What bites is the annual commitment, which turns a cheap experiment into a fixed obligation. Name the person with authority to cancel before anyone signs.
Billing units and unanswered calls
Does Synthflow charge for calls that aren't answered? Get this in writing from Synthflow. Its usage terms define what counts as a billable minute.
Whether ringing time, voicemail pickups and failed attempts count matters most for outbound campaigns, where unanswered dials are the bulk of the volume. Telephony carriers have their own billing rules, so a call can cost nothing on one invoice and something on another.
Is Synthflow priced per minute or per call? Per minute. Usage is metered as duration, so a call is a length of time rather than a unit.
A short call and a long one aren't equivalent, and a flow that keeps callers on the line costs more. Model your average handle time before you model the plan.
Discounts
Does Synthflow offer discounts for agencies or annual billing? Treat any discount as unconfirmed until it appears on a quote. Synthflow's pricing and agency pages show what is published, and anything beyond that is a sales conversation.
Ask what the discount covers: platform fees only, or minutes, telephony and add-ons too. A saving that excludes the largest line item is smaller than it looks.
Related reading
References
Running the numbers
See what the same workload costs here.
Free forever plan, and every paid plan self-hosts at no extra cost.
See pricing Talk to sales
