Disclosure: some links in this article are affiliate links. If you buy through them, Sophrai may earn a commission at no extra cost to you. It never changes our rankings. Read our affiliate disclosure. Sophrai may earn a commission from the Vapi link (/go/vapi).
Short answer: pick Vapi if you’re a developer who wants the lowest raw per-minute cost and full control over the speech, model and voice stack. Pick Retell AI if you want simple pay-as-you-go pricing with 20 concurrent calls included. Pick Bland AI if you want one all-in per-minute price with no token bills. On 30 October 2026, a typical US call costs about $0.08–$0.13/min on Vapi, ~$0.125/min on Retell and $0.14/min plus telephony on Bland Start.
All prices come from each vendor’s official pricing page, checked 30 October 2026. We haven’t benchmarked latency or voice quality.
AI voice agents now answer inbound calls, qualify leads, book appointments and chase invoices for a lot of B2B teams. The three developer platforms most teams shortlist are Vapi, Retell AI and Bland AI. They price in very different ways: Vapi passes model costs through at cost and adds a hosting fee, Retell bundles its infrastructure and lets you choose models per minute, and Bland charges one all-in rate. That makes headline prices hard to compare. This guide puts them on the same basis, shows what you get at each tier, and explains how to wire each one into n8n for CRM updates and follow-ups.
Vapi vs Retell AI vs Bland AI at a glance
| Vapi | Retell AI | Bland AI | |
|---|---|---|---|
| Pricing model | $0.05/min hosting + providers at cost | Per-minute components (infra + LLM + voice + phone) | All-in per-minute rate (LLM, STT, TTS included) |
| Typical per-minute cost (US) | $0.082–$0.129 (Vapi’s estimate) | $0.11 + $0.015 telephony | $0.14 Start / $0.12 Build, plus telephony |
| Monthly platform fee | $0 (Core $29, Pro from $999) | $0 (Enterprise custom) | $0 Start / $299 Build |
| Concurrent calls included | 4 (10 on Core, 30 on Pro) | 20 | 10 Start / 50 Build |
| Extra concurrency | $10 per line/month | $8 per call/month | Move up a plan |
| Daily call cap | Not listed | Not listed | 100 Start / 2,000 Build |
| Free start | $5 credits, 1 Vapi number | Pay-as-you-go from $0 | 2 credits + inbound number, no card |
| HIPAA | $2,000/month add-on (DPA + BAA) | “HIPAA-ready”; BAA tier not confirmed | BAA on Enterprise |
Prices in USD, checked on the official pricing pages on 30 October 2026. “Not listed” means the pricing page didn’t state a limit; it isn’t proof that none exists.
How much does Vapi cost?
Vapi charges a $0.05 per minute hosting fee and passes speech-to-text, the language model, text-to-speech and transport through at the providers’ listed prices, with no markup. With its default stack, Vapi’s own calculator shows Deepgram at $0.0095–$0.0099/min, OpenAI at $0.0077–$0.0452/min and ElevenLabs at $0.0146–$0.0238/min, which is $82–$129 for 1,000 minutes.
Telephony through Vapi Telephony/SIP, WebSockets or Daily WebRTC is listed as free. If you bring Twilio, Vapi lists $0.008/min inbound and $0.014/min outbound (charged by the provider); Telnyx is $0.0055/min and Vonage $0.00814/min.
On top of usage you can add a success package:
- No package: $5 free credits, 2 organizations, 1 Vapi phone number, 4 concurrent calls, 14 days of raw data retention, community and email support.
- Core, $29/month: 5 numbers, 10 concurrent calls, 30-day retention, zero data retention (ZDR), email support with a 2-business-day SLA.
- Pro: 10% of your Vapi hosting fees with a $999/month minimum. 10 numbers, 30 concurrent calls, 180-day retention, RBAC, dedicated Slack, 99% uptime SLA.
- Premier: contact sales, 99.9% uptime SLA, SSO and dedicated SIP.
Add-ons work on any package: extra concurrency is $10 per line a month, extra organizations $20 each, and HIPAA-eligible data handling is $2,000 a month per organization.
How much does Retell AI cost?
Retell is pay-as-you-go with no platform fee, and its pricing page lists $0.07–$0.31 per minute depending on the parts you choose. The core is Retell Voice Infra at $0.055/min. You then add a voice and a model:
- Voices: Retell platform, Cartesia, OpenAI, Minimax, Fish and Inworld voices are $0.015/min; ElevenLabs runs from $0.040/min (Flash) to $0.10/min (v3, v4).
- LLMs: from $0.0016/min (GPT 5 nano) to $0.384/min, with the recommended models at $0.064/min (GPT 5.6 Terra, Claude 5 Sonnet) and $0.08/min (GPT 5.4). Speech-to-speech models such as GPT Realtime mini are $0.07/min.
- Telephony: $0.015/min for Retell’s US Twilio line; no charge if you bring your own SIP trunk.
Retell’s calculator default (an LLM at $0.04/min plus a $0.015 voice) comes to $0.11 per minute before telephony. Monthly items include $2 per Retell phone number, $8 per extra concurrent call above the free 20, and $8 per knowledge base above the free 10. Add-ons include knowledge base retrieval (+$0.005/min), PII removal (+$0.01/min), branded caller ID (+$0.10 per outbound call) and AI quality assurance ($0.10/min after 100 free minutes).
Two billing details are worth knowing. Retell tracks calls to the nearest second and doesn’t round each call up, and once a call is transferred the AI agent fee stops and only telephony continues. Silence and hold time are billed, because speech-to-text keeps listening.
How much does Bland AI cost?
Bland charges one rate that includes the LLM, speech-to-text and text-to-speech, with no token charges:
- Start: $0.14/min, $0 platform fee, no card needed. Transfers $0.05/min, 10 concurrent calls, 100 calls a day, 1 voice, 10 knowledge bases.
- Build: $0.12/min plus a $299/month platform fee. Transfers $0.04/min, 50 concurrent calls, 2,000 calls a day, 5 voices, 50 knowledge bases.
- Enterprise: custom pricing, dedicated infrastructure, on-prem or VPC deployment, BAA, SSO and data residency.
Telephony is extra: bring your own Twilio or SIP trunk, or use Bland’s Twilio at pass-through cost. Bland’s pricing page doesn’t publish that pass-through rate, so we can’t include it. Note that several features many teams expect, including warm and live transfers, guardrails, SMS, appointment scheduling and alarms, are listed as Enterprise-only in Bland’s plan table.
Bland’s pricing FAQ claims production stacks cost “roughly $0.13 to $0.30 per minute on Vapi and $0.11 to $0.25 per minute on Retell”. That’s a competitor’s estimate, not a verified figure.
What do 1,000 and 10,000 minutes a month cost?
Here is the same workload on each platform: US calls, an average of 3 minutes, one phone number and the vendor’s default or mid-range model and voice.
| Monthly minutes | Vapi (no package) | Retell AI | Bland AI |
|---|---|---|---|
| 1,000 (~330 calls) | $82–$129 | ~$127 ($125 usage + $2 number) | $140 + telephony (Start) |
| 10,000 (~3,330 calls) | $820–$1,290 (+$29 if you need Core for 10 lines) | ~$1,252 | $1,499 + telephony (Build: $1,200 + $299) |
Our arithmetic from the official rates on 30 October 2026. Vapi’s range comes from its own calculator for Deepgram + OpenAI + ElevenLabs. Retell assumes its calculator default ($0.11/min) plus $0.015/min Retell Twilio. At 10,000 minutes, Bland Start’s 100-calls-a-day cap (about 3,000 a month) forces the Build plan. Bigger models, ElevenLabs voices or add-ons raise all three.
Three takeaways. First, Vapi is cheapest on paper, but only if you stay on low-cost models and voices, and the 4 default concurrent lines are tight for outbound campaigns. Second, Retell’s 20 free concurrent calls give it the most headroom before you pay for capacity. Third, Bland’s all-in price is easier to forecast, and the difference shrinks if you would otherwise use expensive models on Vapi or Retell.
Which one should you choose?
- Choose Vapi if you have developers, want to pick each provider (STT, LLM, voice, carrier) yourself and want the lowest raw cost. Budget for Core or extra lines once you run more than 4 calls at once, and for the $2,000/month HIPAA add-on if you need it.
- Choose Retell AI if you want transparent per-minute pricing, a no-code agent builder with webhooks and testing, and room to scale to 20 parallel calls without a platform fee. It suits inbound support, booking lines and agencies running several client agents.
- Choose Bland AI if finance wants one predictable per-minute number and you’re happy with Bland’s included models, or if you’ll eventually need on-prem/VPC deployment through Enterprise. Check the Start plan’s 100-calls-a-day cap before an outbound campaign.
If you’re still deciding whether a voice agent is the right channel at all, compare it with text-first options. Our WhatsApp Business API pricing guide covers per-message costs, and Lindy vs Relevance AI vs n8n compares agent platforms that handle email and chat.
How do you connect a voice agent to n8n?
All three platforms can call an n8n workflow during a call (for example, to look up an order or check a calendar) and send the transcript and outcome to n8n after the call. The pattern is the same: an n8n Webhook node receives the request and a Respond to Webhook node sends the answer back to the agent.
- Retell: add a custom function that points to your n8n production webhook URL. Retell sends the function name, the call object (including the transcript so far) and the arguments, and signs the body with an
X-Retell-SignatureHMAC-SHA256 header. For post-call automation, Retell webhooks sendcall_started,call_endedandcall_analyzedevents, time out after 10 seconds and retry up to 3 times, so make your workflow idempotent. n8n has a ready Retell custom function template. - Vapi: Vapi’s integrations docs connect n8n through an API Request tool and a published webhook. Use a Function tool instead if you want Vapi’s full call context; your workflow must return HTTP 200 with a
resultsarray that matches eachtoolCallId. - Bland: create a custom tool with your webhook URL (GET or POST only, 10-second default timeout), and add a
webhookfield to the Send Call request to receive the full call object, including the transcript, when the call ends.
Before you go live, protect those public webhooks: require an auth header or verify the vendor’s signature, and limit what the workflow can write. Our guide to securing your n8n instance lists the settings to check. Run the finished workflow through the free n8n Workflow Auditor, estimate execution costs for post-call workflows in the n8n Cost & Hosting Planner, and compare the AI subscriptions behind your stack in the AI SaaS Pricing Comparison.
What we couldn’t verify
- Latency and voice quality. We didn’t run test calls, so we don’t rank the platforms on how natural they sound or how fast they respond. Run the same script on all three before you commit.
- Bland’s telephony pass-through rate, which isn’t published on its pricing page.
- Which Retell tier includes a BAA. The plan table lists HIPAA/BAA, but the page doesn’t show clearly whether pay-as-you-go includes it.
- Compliance claims (SOC 2, HIPAA, ISO 27001, FedRAMP and so on) are the vendors’ own statements. Ask for reports under NDA.
Frequently asked questions
Which is cheapest: Vapi, Retell AI or Bland AI?
On list prices checked on 30 October 2026, Vapi has the lowest raw per-minute cost: Vapi estimates $0.082 to $0.129 per minute for its default stack (a $0.05 hosting fee plus Deepgram, OpenAI and ElevenLabs at cost), with Vapi telephony free. Retell’s default calculator setup is $0.11 per minute plus $0.015 for Retell’s US Twilio line. Bland Start is $0.14 per minute all-in for the AI, plus telephony. Your real cost depends on the model and voice you choose.
Does Bland AI’s price include the LLM?
Yes. Bland’s pricing page says its per-minute rate covers the LLM, speech-to-text and text-to-speech, with no token charges. Telephony is billed separately, either on your own carrier or on Bland’s built-in Twilio at pass-through cost. Bland doesn’t list that pass-through rate on its pricing page.
How many concurrent calls do I get?
Retell includes 20 concurrent calls on pay-as-you-go, then charges $8 per extra concurrent call per month. Vapi includes 4 concurrent calls without a success package and 10 on Core ($29 a month), with extra lines at $10 each per month. Bland Start allows 10 concurrent calls and 100 calls a day; Build allows 50 concurrent calls and 2,000 a day.
Can I connect Vapi, Retell or Bland to n8n?
Yes, all three can call an n8n Webhook during a call and send post-call data to one. Retell uses custom functions and call webhooks, Vapi uses API Request or Function tools, and Bland uses custom webhook tools plus a post-call webhook. Your n8n webhook must be public, and you should verify signatures or require an auth header.
Is there a free plan?
None of them is free to run at scale, but all three let you start without a contract. Vapi gives $5 in free credits, Bland Start has no platform fee and includes 2 credits plus an inbound number, and Retell is pay-as-you-go from $0 with no minimum. You pay for every connected minute after that.
Which one is HIPAA compliant?
Vapi sells HIPAA-eligible data handling as a $2,000 a month add-on per organization, with a DPA and BAA. Bland says it is HIPAA compliant with a signed BAA, and lists BAA on its Enterprise plan. Retell describes itself as SOC 2 certified and HIPAA-ready, and its plan table lists HIPAA/BAA, but we couldn’t confirm from the page which tier includes a BAA. Ask sales in writing before you handle patient data.
Related reading
- ElevenLabs vs Cartesia: TTS and voice agent pricing compared
- Vapi vs ElevenLabs Agents: voice AI pricing and concurrency
- Zapier vs Make vs n8n AI agents: pricing and control
- Lindy vs Relevance AI vs n8n for B2B AI agents
- Instantly vs Smartlead for cold email
- n8n Cloud vs self-hosted
Sources and method
Prices and limits come from each vendor’s official pricing page, checked on 30 October 2026: Vapi pricing, Retell AI pricing and Bland AI pricing. Integration details come from Retell custom functions, Retell webhooks, Vapi integrations, Vapi API Request vs Function tools, Bland custom tools, Bland agent quickstart, n8n Retell template and n8n Respond to Webhook docs. Monthly totals are our own arithmetic from those rates. Model names and per-minute prices change often, so check the pricing pages before you budget.
