Vapi alternatives: what each one really costs and how to switch without dropping calls
Ten platforms compared on full cost per 1,000 minutes, concurrency and compliance, plus an eight-step migration plan.
The short answer: Retell AI if you want the closest drop-in replacement, with 20 free concurrent calls instead of 4. ElevenLabs Agents if you want a flat monthly bill with minutes included. Bland AI if you want one per-minute price for model, voice and transcription. Telnyx if you handle health data and need HIPAA without a $2,000 add-on. LiveKit or Pipecat if your engineers want out of the platform fee. Twilio ConversationRelay if your numbers already live on Twilio. Goodcall, Voiceflow or Synthflow if you'd rather not build at all.
One thing to know before you pick: Vapi's $0.05 per minute is only the hosting fee. Transcription, the LLM, the voice and telephony stack on top. Vapi's own pricing example puts 1,000 minutes at $82 to $129. We compared the ten alternatives on that full cost, not on the headline rate, and laid out what a real migration involves.
The ranking is editorial: a sponsored spot is always labeled on its card, and no affiliate link ever changes a rating.
Why teams leave Vapi, and what staying costs
Vapi is a developer platform: you pick each piece (speech-to-text, the LLM, text-to-speech, the carrier) and Vapi orchestrates them during the call. That flexibility is the point, and even critics admit it's quick to get started. For the wider market, see our best AI voice agents ranking. People leave over six recurring issues that show up on developer forums and in Vapi's own support threads.
$0.05 a minute is only one line
The headline rate is prepaid Vapi hosting. Models pass through at cost: Deepgram transcription runs about $0.0095 a minute, OpenAI models up to $0.0452, an ElevenLabs voice up to $0.0238. Twilio adds $0.008 inbound and $0.014 outbound. Vapi's own example lands at $82 to $129 for 1,000 minutes. One developer on Hacker News put it bluntly: the platform fee was 60 to 70% of their total spend, and that alone pushed them to self-host.
No support SLA below $29 a month
Without a plan, support means the Discord community and email with no SLA. Core ($29 a month) promises a two-business-day email response. A dedicated Slack channel and a 99% uptime SLA come with Pro, at a $999 monthly minimum. On a public review site, Vapi sits around 2.5 out of 5 on roughly a dozen reviews: a tiny sample, but slow support shows up in nearly every complaint.
Four concurrent calls, then a hard stop
The no-plan tier allows 4 concurrent calls, Core 10, Pro 30. Extra lines cost $10 each per month. According to a support answer, the call over your limit gets rejected, not queued, unless you're on an Enterprise contract. For an inbound line with spikes (a campaign launch, Monday morning), that's a customer hearing a dead line.
Latency depends on the chain you build
Vapi doesn't make the voice or the transcription, so quality and speed depend on the pieces you choose. Support threads report real latency three to four times the dashboard estimate, and non-English agents picking up an English accent. Fixable, but tuning every piece is on you.
A developer tool with rough edges
Vapi's power lives in its API, and non-developers get lost fast. Some engineers have the opposite complaint: too GUI-driven. Its workflow builder was described on Hacker News as having a lot of rough edges, and one reviewer says assistant edits in the dashboard fail to save about half the time.
Compliance is a premium add-on
HIPAA with a BAA costs $2,000 a month on top of your plan. Without a plan, data is retained 14 days, 30 on Core. If you need calls processed only in the EU, the pricing page doesn't promise it: get a written answer and the subprocessor list before you sign.
You're leaving because… go to…
The right Vapi alternative depends on what's pushing you out. Usually one card fits. Each links to the full breakdown below.
20 free concurrent calls and $10 in credits.
You want a predictable monthly billElevenLabs AgentsPlans with minutes included, no ElevenLabs telephony fee.
You're tired of adding up AI line itemsBland AI$0.14 a minute with LLM, STT and TTS included; telephony extra.
You need HIPAA without paying $2,000 a monthTelnyxHIPAA included on every plan.
The platform fee is eating your marginLiveKit or PipecatOpen source, about a cent per agent minute.
Your numbers already live on TwilioConversationRelayBring your own LLM, drop the middle layer.
Vapi alternatives, platform by platform
For each one: why switch to it, who it's for, its limit, how to migrate from Vapi, and what it costs. Prices are the ones listed on each vendor's official pricing page in 2026, in US dollars. Retell, ElevenLabs, Bland and Telnyx are managed platforms; LiveKit, Pipecat and Twilio hand you more control and more code; Goodcall, Voiceflow and Synthflow are for teams that don't want to build.
Retell AI: Vapi's model, with more room to grow
Closest replacementWhy switch. Retell works the way Vapi does: you combine an LLM, a voice and a number, and pay per minute. The difference shows on day one. You get $10 in free credits and 20 free concurrent calls, where Vapi gives you 4. Past that, each concurrent call is $8 a month. The visual flow editor also helps anyone who fought Vapi's workflow builder. In one AI-agent builder community, a member listed why they moved: lower latency, better high-volume pricing and easier hookups to n8n and CRMs.
Who it's for. Agencies and dev teams running agents for several clients who kept hitting Vapi's concurrency cap.
The catch. The bill is still a sum of parts: $0.07 to $0.31 a minute depending on model and voice. Picking ElevenLabs v3 takes the voice line alone from $0.015 to $0.10 a minute. Retell is SOC 2 certified and HIPAA-ready, but the BAA and a custom DPA are Enterprise items.
Migrating from Vapi. Copy your system prompt, first message and tool list. Each Vapi tool that hits your API becomes a custom function in Retell, pointed at the same endpoint. If you automate with Make, the Retell AI Make integration replaces the Vapi one.
Pricing. Run the math on a common setup: infrastructure $0.055, telephony $0.015, platform voice $0.015 and GPT 5.4 mini $0.024. That's $0.109 a minute, or $109 per 1,000 minutes. Swap in GPT 5.4 nano ($0.0064) and you're at $91.40. Add-ons: knowledge base +$0.005 a minute, PII redaction +$0.01, QA at $0.10 a minute after 100 free minutes, branded caller ID +$0.10 per outbound call.
- Free
- $10 credits, 20 concurrent calls
- Price
- $0.07 to $0.31/min
- Best for
- agencies, dev teams
- Catch
- itemized bill, DPA on Enterprise
ElevenLabs Agents: the benchmark voice, on a flat plan
Predictable billWhy switch. Plenty of Vapi agents already run an ElevenLabs voice. Going straight to the voice vendor removes a layer and changes how you pay: a monthly plan with minutes included instead of a per-minute sum. Pro is $99 a month for 1,238 minutes and 20 concurrent calls. ElevenLabs adds no telephony fee: you connect your own carrier (Twilio or a SIP trunk), which bills you directly for the call. Our full ElevenLabs review rates it 4.2 out of 5.
Who it's for. Companies with steady volume that want the best-sounding voice and a budget they know in advance.
The catch. LLM usage is billed separately on top. Extra minutes cost $0.08, and calls over your concurrency limit run $0.16 a minute. Budget your carrier's bill for the phone side of each call. A HIPAA BAA is Enterprise only.
Migrating from Vapi. Keep the same voice ID: it's the one piece that doesn't change, so callers hear the same agent. Port the prompt, then rebuild tools as calls to your API. Start on Creator ($22 for 275 minutes) while you test. Still unsure? Our ElevenLabs alternatives page covers its limits.
Pricing. Free for 15 minutes and 4 concurrent calls. Starter $6 (75 minutes), Creator $22 (275), Pro $99 (1,238), Scale $299 (3,738), Business $990 (12,375), Enterprise on quote. Pro works out to about $0.08 a minute before the LLM.
- Free
- 15 min, 4 concurrent calls
- Paid
- Pro $99/mo for 1,238 min
- Best for
- steady volume, voice quality
- Catch
- LLM billed on top
Bland AI: one price per minute, no token math
One AI priceWhy switch. Bland's connected minute includes the LLM, transcription and voice, with no token charges. Telephony is billed separately, on your own carrier or Bland's at pass-through cost. If your Vapi invoice needs a spreadsheet to decode, this is the calmest switch. Start has no platform fee and no card required.
Who it's for. Outbound-heavy teams (follow-ups, lead qualification) that want a unit cost they can explain to finance in one line.
The catch. $0.14 a minute is above the low end of a tuned Vapi stack. Start also caps you at 10 concurrent calls and 100 calls a day. BAA and data residency are listed only on Enterprise. On the plus side, its compliance is the best documented of the group: SOC 2 Type I and II, GDPR with a DPA, ISO 27001.
Migrating from Vapi. Port the prompt and your transfer rules. Transfers to a human are billed separately ($0.05 a minute on Start), so count them if your agent hands off often. The Bland Make integration keeps your scenarios running.
Pricing. Start: $0.14 a minute, so $140 per 1,000 minutes plus telephony. Build: $299 a month plus $0.12 a minute, with 50 concurrent calls and 2,000 calls a day. Do the math first: Build only beats Start past 14,950 minutes a month.
- Free
- $0 platform fee, no card
- Price
- $0.14/min AI included, telephony extra
- Best for
- high-volume outbound
- Catch
- 100 calls/day on Start
Telnyx Voice AI: the carrier that runs the agent too
Carrier plus AIWhy switch. Telnyx is a telecom carrier first. You may already use its numbers inside Vapi, where its telephony costs $0.0055 a minute. Its voice engine (hosted orchestration, STT and TTS) is $0.05 a minute, and the page puts a production agent near $0.06. The headline for regulated teams: HIPAA is included on every plan, and pay as you go goes up to 500 concurrent calls. The Telnyx Make integration keeps your scenarios.
Who it's for. Clinics, health-tech and insurance teams, plus anyone who wants carrier and AI on one invoice.
The catch. The LLM and telephony are still extra. Going live needs a card on file, and no free credits are mentioned. Its security page lists SOC 2 Type II, ISO 27001, PCI DSS and GDPR, but EU data residency isn't described.
Migrating from Vapi. If your numbers are already on Telnyx, this is the shortest move: the number stays put, only what answers it changes.
Pricing. $0.05 a minute for the voice engine, so $50 per 1,000 minutes before LLM and telephony. Usage-based, no subscription.
- Price
- $0.05/min voice engine
- HIPAA
- included on every plan
- Best for
- healthcare, Telnyx numbers
- Catch
- card on file, no credits
LiveKit Agents: get out from under the platform fee
Open sourceWhy switch. It answers Vapi's biggest complaint. The framework and server are open source, so you can self-host, and the cloud version stays cheap. Build is free with no card: 1,000 agent session minutes, a free US number and $2.50 in inference credits. The pricing page's full-call example comes to $0.0479 a minute.
Who it's for. High-volume engineering teams where platform margin outweighs build time.
The catch. Vapi's comfort is gone: you write the agent. The same Hacker News developer who left Vapi over fees said that even on Pipecat and LiveKit, the plumbing was still painful. SOC 2 Type II reports and HIPAA start at Scale ($500 a month).
Migrating from Vapi. Treat it as a build project, not a one-week switch. Keep Vapi in production while the LiveKit agent goes through testing.
Pricing. Build $0/month (1,000 minutes), Ship from $50 (5,000 minutes), Scale from $500 (50,000 minutes), then $0.01 a minute. Models are extra.
- Free
- 1,000 agent min, no card
- Paid
- Ship from $50/mo
- Best for
- high volume, in-house devs
- Catch
- you code everything
Pipecat Cloud: the open-source framework, hosted
Cheapest per minuteWhy switch. Pipecat is an open-source voice agent framework, and Pipecat Cloud (from Daily) hosts it for $0.01 per active agent minute, with Krisp noise cancellation included up to 10,000 minutes. A Pipecat maintainer said it honestly on Hacker News: the integrated developer experience is better on Vapi, and Pipecat is for teams that want to control every part.
Who it's for. Developers who want every latency and cost knob in their own hands.
The catch. The pricing page mentions neither HIPAA nor GDPR. As with LiveKit, you build the agent yourself. Pieces add up: real-time transport at $0.018 a minute, SIP at $0.003 to $0.02.
Migrating from Vapi. Keep your model, voice and STT choices; Pipecat lets you pick each one. Prompts and tools get rewritten as code.
Pricing. $0.01 per active agent minute, so $10 per 1,000 minutes, plus $0.0005 per reserved minute. Models and telephony extra.
- Price
- $0.01/active agent min
- Included
- Krisp up to 10,000 min
- Best for
- hands-on developers
- Catch
- no HIPAA or GDPR info
Twilio ConversationRelay: keep Twilio, bring your own LLM
Already on TwilioWhy switch. If your numbers are on Twilio, Vapi is one more layer between your carrier and your model. ConversationRelay handles STT and TTS, detects the caller's language, and is built for bring-your-own-LLM. One less vendor.
Who it's for. Existing Twilio customers with a developer who can wire up an LLM.
The catch. No free tier is described. The LLM is on you, and everything is built in code.
Migrating from Vapi. Your numbers don't move: you change what each one points to in the Twilio console. Test on a secondary number first.
Pricing. $0.07 a minute for ConversationRelay, plus the call: $0.0085 inbound and $0.0140 outbound in the US, and $1.15 a month for a local number. For 1,000 inbound minutes: $78.50, or $79.65 with the local number, before the LLM.
- Price
- $0.07/min plus the call
- US number
- $1.15/mo
- Best for
- Twilio customers
- Catch
- no free tier
Goodcall: a flat bill per customer, not per minute
No per-minute billingWhy switch. Goodcall is a turnkey AI answering service for small businesses, and it bills in a way no one else on this page does: per unique caller, never per minute. A thirty-second wrong number and a nine-minute booking cost the same. If Vapi's per-minute stack left you unable to predict your bill, this is the opposite model. You can start free with a demo agent and add a number when you're ready.
Who it's for. A solo tradesperson or a busy shop that misses calls on the job and has no developer, the kind of business that would never build on Vapi's API.
The catch. It's a receptionist, not a platform. Starter allows 1 external tool and 30 days of call history; Growth gets 5 tools, Scale 15. Custom Vapi tools and squads won't carry over one for one.
Migrating from Vapi. Rewrite your Vapi prompt as plain business instructions, keep the one or two integrations that matter (usually the calendar), and point your number at Goodcall.
Pricing. Starter $79 a month for 100 unique customers, then 79 cents each (about 150 calls). Growth $129 for 500, then 26 cents. Scale $299 for 2,000, then 15 cents. Enterprise on quote.
- Free
- demo agent to start
- Paid
- Starter $79/mo, 100 customers
- Best for
- small service businesses
- Catch
- 1 external tool on Starter
Voiceflow: one builder for voice and chat agents
For agenciesWhy switch. Voiceflow builds AI agents for voice and chat from the same workspace, which Vapi doesn't do. Agencies get a free trial with no credit card, usage-based billing, multi-client workspaces and white-labeling, with access to all the major model providers or your own model. Its footer lists SOC 2, ISO 27001, GDPR and HIPAA.
Who it's for. Agencies that ship agents for several clients and want one tool for the phone line and the website chat.
The catch. Business pricing is on request, so you can't compare it per minute before a sales call. It's a CX platform first: for a pure outbound dialer, Retell or Bland fit better.
Migrating from Vapi. Rebuild the conversation as a workflow plus global agent instructions, then reconnect your APIs as integrations. Use the free trial to rebuild one Vapi assistant before you commit.
Pricing. Agencies: free trial, then usage-based billing. Businesses: pricing on request, after a demo.
- Free
- trial for agencies, no card
- Paid
- usage-based, business on request
- Best for
- agencies, voice plus chat
- Catch
- no public business price
Synthflow: the managed route for large call volumes
Enterprise contractWhy switch. Synthflow appears in almost every Vapi alternatives list, and for a reason: it sells a scoped deployment rather than a toolkit. Call volume, phone system, integrations, security review and launch support go into one annual contract. It covers SOC 2 Type 2, ISO/IEC 27001, HIPAA and GDPR, and BPOs and software platforms can add white-label in that contract.
Who it's for. Contact centers and larger companies that would rather pay for a managed launch than staff a voice-agent team.
The catch. There is no free plan, no trial and no rate by the minute. You can't price it against Vapi without a scoping call, which is why it ranks last here despite its reputation.
Migrating from Vapi. Bring your Vapi call volume, your carrier setup and your security requirements to the scoping call; the team prices the deployment from those.
Pricing. Annual enterprise agreement, quoted on your volume and requirements.
- Price
- annual contract, quoted
- Compliance
- SOC 2 Type 2, HIPAA, GDPR
- Best for
- contact centers
- Catch
- no free plan or trial
Vapi vs the 10 alternatives, side by side
To compare like for like, take 1,000 minutes a month. Vapi: $82 to $129 by its own example. Retell with GPT 5.4 mini: $109. Bland Start: $140 for the AI, telephony extra. ElevenLabs Pro: $99 for 1,238 minutes, LLM extra. Twilio ConversationRelay inbound: $78.50 before the LLM. LiveKit Build: $0 in platform fees. Goodcall Starter: $79 a month, billed per caller, not per minute. The gaps come from what's included, not from the headline rate.
| Entry price | Free tier, concurrency, compliance | |
|---|---|---|
| VapiWhere you start | $0.05/min hosting + models + telephony; Core $29/mo | $5 credits, 4 concurrent calls; HIPAA $2,000/mo |
| Retell AIClosest swap | $0.07 to $0.31/min depending on stack | $10 credits, 20 concurrent calls; SOC 2, HIPAA-ready (BAA on Enterprise) |
| ElevenLabs AgentsFlat plans | Pro $99/mo for 1,238 min; $0.08/min extra | 15 free min, 4 concurrent calls; BAA on Enterprise |
| Bland AIAI included | $0.14/min on Start; Build $299/mo + $0.12/min | No card, 10 concurrent calls; SOC 2, GDPR, ISO 27001 |
| TelnyxCarrier | $0.05/min voice engine | No credits; HIPAA on every plan, SOC 2 Type II |
| LiveKit AgentsOpen source | Build $0, Ship from $50/mo | 1,000 free min; HIPAA from Scale ($500/mo) |
| Pipecat CloudHosted framework | $0.01/active agent min | Krisp included up to 10,000 min |
| Twilio ConversationRelayTwilio users | $0.07/min + call ($0.0085 inbound) | No free tier |
| GoodcallPer caller | Starter $79/mo for 100 customers | Demo agent free; 1 tool on Starter |
| VoiceflowVoice and chat | Usage-based; business on request | Agency trial with no card; SOC 2, ISO 27001, HIPAA |
| SynthflowEnterprise | Annual contract, quoted | No free plan or trial; SOC 2 Type 2, HIPAA, GDPR |
Prices from official vendor pages in 2026, before tax and subject to change. Most per-minute rates exclude the LLM: rerun the math on your own volume.
Switch from Vapi in eight steps
Order matters. A voice agent that answers badly costs more than one that doesn't answer at all, so each step protects live calls.
Inventory everything running on Vapi
Day 1List every assistant, every squad (Vapi's multi-assistant handoff that keeps context), every tool, phone number and server webhook. Add the Make or n8n scenarios hooked into them (each Vapi Make module needs a replacement). A forgotten tool is a booking that silently stops reaching a client's calendar.
- Time
- one hour to half a day
- If rushed
- a missed tool or scenario
Work out your real cost per minute
DecisionTake your last invoice and split it three ways: Vapi hosting, models (STT, LLM, TTS) and telephony. Divide by minutes. That number, not $0.05, is what you compare. One user paying about $0.20 a minute found the gap came mostly from Twilio telephony billed on top.
- Vapi benchmark
- $82 to $129 per 1,000 min
Back up prompts, voices and settings
Day 1Copy every system prompt, first message, voice ID, model, transcription language and end-of-call rule into a doc. Without a plan, Vapi keeps data for 14 days: download the recordings and transcripts you'll need to test the new platform.
- Vapi retention
- 14 days free, 30 on Core
Rebuild tools and webhooks
BuildEach Vapi tool calls an API: recreate the same call at the destination (a custom function on Retell, a server tool elsewhere). Keep the same endpoints on your side and add a field naming the source platform, so you can tell traffic apart while both run.
- Tip
- a source field on every request
Rebuild the handoffs
BuildIf you use Vapi squads, rebuild the handoffs in the destination's flow. Two or three simple assistants often fit in one agent with a better-structured prompt. Fewer handoffs, less latency.
- Gain
- fewer transfers, lower latency
Check who owns each number
CarefulA number bought on Twilio, Telnyx or Vonage stays with that carrier; you only change what answers it. A number provisioned by Vapi needs a conversation with Vapi support before you cancel. Get the answer in writing first.
- If rushed
- a lost number, callers into the void
Run both platforms side by side
One to two weeksPoint a secondary number at the new platform, then a slice of live traffic. Listen to real calls: names, addresses, numbers read aloud, silences. Measure latency against the estimate, since Vapi users have reported three to four times the dashboard figure. Check your peak too: how many concurrent calls go through without a rejection.
- Measure
- latency, accuracy, rejected calls
Cut over, then stop auto-renew
Last dayMove the main number on a quiet day. Keep Vapi open long enough to pull the last call logs. Core is prepaid with auto-renew: cancel before the date or you're paying for another month.
- Trap
- Core auto-renewal
Is your tool missing from this page? Email us: hello@hackceleration.com
What makes a Vapi migration go wrong
The same mistakes come up again and again in voice-agent developer threads. None is deeply technical. They're oversights, and callers hear every one of them.
Comparing half a price
Vapi's $0.05, Telnyx's $0.05 and Retell's $0.055 infrastructure line don't cover the same thing. Always compare a full cost for 1,000 minutes, LLM and telephony included.
Demo-testing instead of call-testing
A scripted demo always sounds great. Accent drift, misheard names and garbled phone numbers only show up on real calls with real callers. Test with your actual scripts.
Sizing for average, not peak
4 concurrent calls on Vapi's free tier, 10 on Bland Start, 20 on Retell and ElevenLabs Pro. Size the plan to your busiest hour, not your average.
Losing your recordings
Without a plan, Vapi deletes after 14 days. Those recordings are your best test set for the new platform: download them first.
Canceling before the numbers are sorted
A platform-provisioned number isn't always portable. Settle number ownership in writing before you cancel anything.
Assuming compliance
On Bland, BAA and data residency are Enterprise. On ElevenLabs, the BAA is too. Retell is HIPAA-ready but the BAA and custom DPA are Enterprise. Only Telnyx includes HIPAA on every plan. Get the paperwork before cutover.
- Assistants, squads, tools, numbers, webhooks and scenarios listed
- Real cost per minute worked out from an actual Vapi invoice
- Prompts, voice IDs and settings copied out of Vapi
- Recordings and transcripts downloaded before deletion
- Ownership of every number confirmed in writing
- Tools rebuilt against the same server endpoints
- Parallel run on real calls
- Peak concurrency checked on the new platform
- Vapi auto-renew canceled after cutover
Vapi alternatives: frequently asked questions
What does Vapi actually cost per minute?
About $0.08 to $0.13 a minute in Vapi's own example, or $82 to $129 per 1,000 minutes. The $0.05 headline covers hosting only; transcription, the LLM, the voice and telephony are added. Core adds $29 a month. On Twilio, count $0.008 inbound and $0.014 outbound. Vapi's own telephony, SIP and WebRTC are free, which is the first lever to cut the bill without switching.
Does Vapi have a free plan?
Yes, a small one. You start at $0 with $5 in credits, 4 concurrent calls, 1 phone number, 2 orgs and 14-day data retention. Support is Discord plus email with no SLA. For a longer trial elsewhere, Retell gives $10 and 20 concurrent calls, LiveKit 1,000 agent minutes a month, Bland a no-card start.
What's the best no-code alternative to Vapi?
Goodcall, if you run a small business: $79 a month on Starter, billed per caller, with a free demo agent. Voiceflow suits agencies (free trial, no card). Synthflow fits large teams, but only as an annual contract with no free plan or trial. In Europe, Fonio and Famulor sell turnkey phone agents billed in euros. If you mostly need a business phone system with some AI, see our Aircall review.
Is there an open-source alternative to Vapi?
Yes, two serious ones: LiveKit Agents and Pipecat. Both run on your servers or in their clouds, at about $0.01 per agent minute beyond plan allowances. The price is engineering time: you write the agent in code. Vocode, once a common pick, no longer offers a public hosted product, just its code on GitHub.
Which Vapi alternative includes HIPAA without the $2,000 add-on?
Telnyx, which includes HIPAA on every plan, with a voice engine at $0.05 a minute. Retell calls itself HIPAA-ready (BAA on Enterprise), and LiveKit offers HIPAA from its Scale plan ($500 a month). On Bland and ElevenLabs, the BAA is an Enterprise item, priced on quote.
Can I keep my providers (ElevenLabs, Twilio) after leaving Vapi?
Yes, on platforms that let you assemble pieces. Retell offers ElevenLabs v3 at $0.10 a minute for the voice. ConversationRelay runs on your Twilio numbers. ElevenLabs Agents adds no telephony fee, but you bring your own carrier (Twilio or SIP), which bills you for the call. Single-price tools like Bland pick the AI stack for you.
Retell vs Vapi: which is cheaper?
On an identical stack, the gap is small: $0.05 hosting on Vapi, $0.055 infrastructure on Retell, models paid on both. The real difference is concurrency: 20 free concurrent calls on Retell versus 4 on Vapi, then $8 per extra line instead of $10. For an agent that takes call spikes, Retell ends up cheaper.
Is Vapi still the right choice for some teams?
Yes. If you're a small dev team prototyping, Vapi's integrated developer experience is still praised, even by people who build on its rivals. Free Vapi telephony, SIP and WebRTC keep costs down on internal or web-based agents. Leave when concurrency, support or the platform fee start costing you calls or margin.
Get the next migration guide in your inbox
One tool to leave, one destination, one migration plan. Nothing else.