Quick answer: Vapi is a hosted voice-agent platform you configure; Pipecat is Daily’s open-source Python framework (BSD-2-Clause) that you code, then run yourself or on Pipecat Cloud. At 10,000 minutes a month on the same Deepgram, OpenAI and Cartesia models, we itemise $770.60–$1,210.10 on Vapi and $363.20–$802.70 on Pipecat Cloud, before carrier charges and engineering time (list prices as at 18 November 2026).
The gap is the platform fee: $0.05 a minute on Vapi against $0.01 an active minute on Pipecat Cloud. On Vapi that money buys you out of running a Python audio service. Telephony, turn detection and observability split the same way: settings on Vapi, code you compose on Pipecat, from Twilio serializers to Daily’s PSTN and the open Smart Turn model.
Where Zian fits: an AI agents platform, not a developer framework
If you want answered calls and booked appointments rather than a voice stack to maintain, look at Zian AI first; LeadsNow.ai is the done-for-you route. Zian AI is a standalone AI sales and reception agents platform, not a phone system (carrier / CPaaS / contact centre) and not an add-on to one, and it is not a developer framework like Pipecat or a build-it-yourself platform like Vapi.
Zian answers inbound calls 24/7 as an AI receptionist that qualifies, books, transfers and takes messages; runs outbound calls; and follows up by SMS, email, WhatsApp, Instagram DMs and Facebook Messenger, with answering-machine and voicemail handling on calls. It supports 30+ languages and voice cloning, connects to HubSpot, Salesforce, HighLevel and Zapier, and offers private model deployment on customer infrastructure. Zian has set 64,521+ qualified sales appointments. LeadsNow.ai runs on Zian as a managed, pay-per-result service that answers leads in under 60 seconds, 24/7, and has set 50,769+ appointments. Zian is in a partnership-application beta with no published prices; each of its three listed plans, including Intel + Widgets (Free Forever), starts with a partnership application.
Want the agents without writing a pipeline or a webhook server? Apply For Partnership
Vapi vs Pipecat: what you build and what you run on each
On Vapi, an assistant is a configuration (transcriber, model, voice, speaking behaviour, tools) set in the dashboard or API, and Vapi runs the call. Its FAQ adds: “For more advanced features like function calling, you will have to set up a Server URL to receive and respond to messages.” Your code is that webhook server; the media pipeline is Vapi’s, and the same FAQ says “No, Vapi does not support on-premise deployments.”
On Pipecat, the agent is a Python program: a framework that “orchestrates AI services (STT, LLM, TTS, and more) into real-time pipelines”, which Pipecat’s introduction says is “free to use under the BSD-2 license, and works with any AI provider and any hosting environment — you pay only for the services you choose.” Pipecat is built and maintained by Daily; the latest PyPI release is pipecat-ai 1.16.0, published 12 November 2026 (as at 18 November 2026). It runs on your own servers, on Pipecat Cloud (Daily-hosted regions, billed per agent minute), or on Pipecat Enterprise, where agents run in a Kubernetes cluster you operate and Pipecat Cloud stays the control plane.
| Question | Vapi | Pipecat (framework + Pipecat Cloud) |
|---|---|---|
| What the agent is | An assistant configuration plus your webhook server for tools | A Python pipeline in your repository, packaged as a container image |
| Licence | Proprietary hosted service | Framework BSD-2-Clause; Pipecat Cloud is a paid hosted service |
| How a change ships | Edit the assistant in the dashboard or API; Vapi’s 12 October 2026 release added percentage traffic splitting across published versions | Code change, image build, pipecat cloud deploy (or deploy from a GitHub branch through the Pipecat Cloud GitHub App) |
| Where it runs | Vapi only; no on-premise option | Your own infrastructure; Pipecat Cloud regions in the US, Europe and India; or Pipecat Enterprise in your own Kubernetes cluster |
| Who scales it | Vapi, within your concurrent-line allowance | Pipecat Cloud autoscaler within your min-agents and max-agents settings, plus a free auto-scaling buffer; on your own servers, you |
| Outbound campaigns | Packaged: Outbound Campaigns (CSV or manual contacts, up to 10,000 per campaign, a schedule for when dialling starts and stops, and a concurrency cap) | No campaign product found in Pipecat’s docs index, README or Pipecat Cloud pricing page, as at 18 November 2026; per-call dial-out and a VoicemailDetector you wire in |
| Paid help and service levels | Core: two-business-day response SLA; Pro: dedicated Slack channel, one-business-day response and 99% uptime SLAs; Premier: named support team (the pricing page names an Agent Strategist and a Forward Deployed Engineer Advisor), one-hour P0 response and 99.9% uptime SLAs | Daily sells enterprise support, including Forward Deployed Engineering engagements; Pipecat Cloud’s Enterprise tier lists 24/7 priority support with dedicated account management |
On Vapi you own a webhook and a prompt; on Pipecat, a long-running audio service, which we priced in engineer-days in our estimate of how long it takes to build a voice AI agent in-house. Choosing between two open-source frameworks instead? Our architectural comparison of LiveKit Agents and Pipecat covers that decision.
How Vapi and Pipecat Cloud charge for a minute
Vapi’s pricing page states the formula plainly: “Usage is your Vapi hosting fee ($0.05/min) plus the model costs for each minute your agents run.” The per-minute cost has “Four components: speech-to-text, the language model, text-to-speech, and transport. Each is billed at the provider’s listed price with no markup.” Usage only is pre-paid with $5 of free credits; packages add Core at $29 a month, Pro at 10% of the hosting fee with a $999 monthly minimum, and Premier by quote. Usage only includes 4 concurrent calls, Core 10 and Pro 30; each extra line is $10 a month. Vapi’s FAQ says volume pricing is available as you scale; we use list prices. We could not find a billing increment (per-second or rounded minutes) on Vapi’s pricing page, billing documentation, FAQ or release notes from June to 9 November 2026, as at 18 November 2026.
Pipecat Cloud bills agent hosting and transport, and sends model costs to the provider. The voice profile, agent-1x (0.5 vCPU, 1 GB memory), costs $0.01 a minute active and $0.0005 a minute reserved; the larger agent-2x and agent-3x profiles cost $0.02 and $0.03 active. Reserved agents are warm instances set with min-agents; billing for them “stops only when the reserved agent is turned off”, and the scaling guide says one “accrues 1,440 reserved minutes a day, even with zero active sessions.” Both are “measured to the second and billed in minutes”, and an active session runs from the /start request until your pipeline shuts down, so it can run slightly longer than talk time. LLM, speech-to-text and text-to-speech are “Billed to provider”: you hold those accounts yourself. Daily WebRTC voice is free for one-to-one sessions, a WebSocket transport is billed by its provider, and Krisp VIVA noise isolation is free to 10,000 active minutes a month, then $0.0015 a minute. Pipecat Cloud’s Enterprise tier also lists “Custom pricing and higher rate limits for high-volume deployments”.
What do 10,000 minutes a month cost on Vapi vs Pipecat?
A1: 10,000 billed minutes in a 30-day month. A2: a peak of 10 simultaneous calls. A3: the same three model vendors on both sides, Deepgram for speech-to-text, OpenAI for the language model and Cartesia for the voice, at the per-minute ranges Vapi’s calculator publishes, which Vapi says are provider prices with no markup; the model rows are identical by construction, and your invoices will differ. A4: calls arrive through Twilio, which bills its own minutes on both routes, so the carrier is left out of both totals. A5: on Pipecat Cloud, agent-1x with one warm agent (min-agents 1); callers beyond it and the free buffer wait for a cold start, “Around 10 seconds” at best. Krisp VIVA, recordings and HIPAA are off. A6: on Vapi, Core at $29 covers 10 lines, which is cheaper than Usage only plus six $10 lines ($60). A7: list prices, no volume discount on either side.
| Line item at 10,000 min/month | Vapi (Core) | Pipecat Cloud (agent-1x) |
|---|---|---|
| Package or plan fee | $29 | $0 (no plan fee listed) |
| Platform minutes | 10,000 × $0.05 = $500 | 10,000 × $0.01 active = $100 |
| Warm capacity | Included in the 10 lines | 1 reserved agent × 43,200 min × $0.0005 = $21.60 |
| Transport | $0 (Twilio billed by Twilio) | $0 (Twilio WebSocket billed by Twilio) |
| Speech-to-text (Deepgram, $0.00953–$0.00994/min) | $95.30–$99.40 | $95.30–$99.40, paid to Deepgram |
| Language model (OpenAI, $0.00774–$0.04523/min) | $77.40–$452.30 | $77.40–$452.30, paid to OpenAI |
| Voice (Cartesia, $0.00689–$0.01294/min) | $68.90–$129.40 | $68.90–$129.40, paid to Cartesia |
| Total, before carrier | $770.60–$1,210.10 | $363.20–$802.70 |
| Effective cost per minute | $0.0771–$0.1210 | $0.0363–$0.0803 |
One caveat: Vapi’s cost guide says it transcribes “audio from both the caller and the assistant. The estimate therefore includes two audio channels.” If your Pipecat pipeline transcribes only the caller’s channel, its Deepgram line would be roughly half of the figure above (our arithmetic on Vapi’s stated assumption).
At 100,000 minutes a month, a 40-call peak and five warm agents on Pipecat Cloud, Vapi platform fees are $5,329 ($5,000 hosting plus Core and 30 extra lines at $329, which beats Usage only plus 36 lines at $360 and Pro at $1,099) and Pipecat Cloud’s are $1,108 ($1,000 active plus 5 × 43,200 × $0.0005 = $108 reserved). With the model rows scaled tenfold, the totals are $7,745–$12,140 on Vapi and $3,524–$7,919 on Pipecat Cloud, before carrier charges.
Rather not price, build or run either stack? Zian runs AI receptionist, outbound calling and messaging agents on one platform. Apply For Partnership
The Four-Cent Test: when Pipecat’s lower hosting fee pays for itself
The Four-Cent Test: multiply your monthly minutes by $0.04. That is roughly what Pipecat Cloud saves you over Vapi on platform fees at list price, with the same models and carrier. If owning the Pipecat code costs your team more than that each month, Vapi is cheaper in total.
The four cents is Vapi’s $0.05 minus Pipecat Cloud’s $0.01. At 10,000 minutes the exact platform gap is $407.40 ($529 against $121.60), and at 100,000 minutes it is $4,221 ($5,329 against $1,108). Telephony can shrink it, as the next section shows, and so can a negotiated volume price on either side.
Now put engineering beside it. Our platform-versus-framework crossover calculation assumes US$1,000 per engineer-day, 3 engineer-days a month of evaluation and upgrade work, 2 of on-call, and an 83-day build amortised over 24 months ($3,458 a month). Those are our own estimates, not measurements, loaded on the Pipecat side only. On them, $407.40 a month buys under half an engineer-day. Holding the 100,000-minute package, line and reserved fees constant, the platform gap covers the ongoing $5,000 a month at about 119,000 minutes, and that plus the amortised build ($8,458) at about 206,000. Below about 119,000 minutes Vapi costs less in total; above about 206,000 Pipecat Cloud does; in between, it depends on whether you count the build. Replace the engineer-day cost with your own first.
Phone numbers, SIP and Twilio on Vapi and Pipecat
Vapi gives you one free US number on Usage only (5 on Core, 10 on Pro), and its documentation is explicit about the limits: “Free Vapi numbers are inbound only. Import a number from a supported telephony provider to make outbound calls.” Free numbers cannot run Outbound Campaigns and cannot be ported out. For outbound, you import a Twilio, Telnyx or DIDWW number, or bring a SIP trunk: Vapi’s docs index has guides for Twilio, Telnyx, DIDWW, didlogic, Bitcall, Zadarma, Plivo and Amazon Chime SDK, and its SIP trunking page describes a custom “BYO” trunk for any provider. Vapi’s pricing page lists Vapi Telephony/SIP, Vapi WebSockets and Daily WebRTC as free, and Twilio at $0.008 a minute inbound and $0.014 outbound, Vonage at $0.00814 and Telnyx at $0.0055, as “Charged by provider, not Vapi”.
Pipecat’s telephony overview lists three connection types: WebSocket streaming from a carrier (Twilio, Telnyx, Plivo, Exotel), WebRTC through Daily’s PSTN service, and SIP. Pipecat Cloud has WebSocket endpoints for those four carriers plus a generic one, and the docs index lists 15 frame serializers, including Bandwidth, Vonage, Genesys AudioHook and FreeSWITCH. On price, the Pipecat Cloud pricing page lists Daily PSTN dial-in and dial-out at $0.018 a minute, Daily SIP at $0.003–$0.02 a minute, SIP REFER transfers at $0.20 per event, and Twilio as “Billed to provider”. The same page adds two conditions: transcoding “results in additional charges appearing on your monthly invoice, including a higher per minute rate”, and a transfer keeps the call “anchored to Daily, so minutes continue to accrue for the duration of the transfer”, unless SIP REFER hands it off. Daily’s Video SDK pricing page shows PSTN at $0.018–$0.03 (“Rates depend on volume and usage of features like transcoding”) and a dial-in number at $2 a month, a fee the Pipecat Cloud page does not list.
| Telephony at 10,000 min/month | Vapi | Pipecat Cloud |
|---|---|---|
| Inbound only, US number from the platform | Free Vapi number; transport listed as free: $0 | Daily PSTN: 10,000 × $0.018 + $2 number = $182 (up to $302 at Daily’s $0.03 upper rate) |
| Outbound, platform-supplied number | Not available: free numbers are inbound only | Daily PSTN dial-out: $182 to $302 on the same basis (dial-out to phone numbers is a paid Daily feature, enabled per domain through a request form; international dial-out through sales) |
| Twilio number, either direction | Billed by Twilio (Vapi lists $0.008 in / $0.014 out) | Billed by Twilio over Media Streams WebSocket |
| Warm or cold transfer | Transfer Call tool: blind transfer by default, plus warm modes (the message, summary and TwiML modes are Twilio only), to a phone number or SIP URI | Daily PSTN cold and warm transfer examples; minutes accrue during an anchored transfer; SIP REFER $0.20 per event |
That first row matters for receptionist-style agents. On an inbound-only US line using each platform’s own number, Pipecat Cloud’s $182 telephony line cuts the $407.40 platform gap to $225.40 at 10,000 minutes, and to $105.40 at Daily’s upper rate.
Transports: phone, WebRTC, WebSocket and WhatsApp
Vapi lists three transports, Vapi Telephony/SIP, Vapi WebSockets (raw PCM or mu-law audio) and Daily WebRTC. Pipecat’s README lists Daily, LiveKit, SmallWebRTC and Vonage WebRTC transports, FastAPI and server WebSockets, WhatsApp and Local, and Pipecat Cloud documents WhatsApp Business voice calls through the WhatsApp Cloud API and SmallWebRTC. If you would rather host on LiveKit’s own cloud, our Vapi vs LiveKit Agents cost comparison prices that route.
Turn detection: Vapi’s speaking plans vs Pipecat’s Smart Turn
Vapi exposes turn detection as two settings. The Start Speaking Plan waits 0.4 seconds by default; smart endpointing can use LiveKit’s model (Vapi recommends it for English), Vapi’s own (recommended for non-English), the built-in turn detection of Deepgram Flux or AssemblyAI, or a custom endpointing model on your server that returns a timeout from 0 to 15 seconds. The Stop Speaking Plan sets how many words interrupt the assistant (0 for an immediate reaction), how long the caller must speak first (default 0.2 seconds) and the pause before it resumes (default 1 second).
Pipecat makes turn detection a pipeline strategy. Since v0.0.102 the default stop strategy uses LocalSmartTurnAnalyzerV3, the open Smart Turn model, which runs in-process on CPU with weights bundled in the package; Pipecat Cloud’s guide calls it “Free to use: The model is fully open source”. Alternatives include turn-1-mini, Krisp VIVA Turn and Alkove Turn, plus start strategies including VAD, a received transcription, a minimum word count, a wake phrase and Krisp’s interruption-prediction model: more control, and more tuning to own.
Observability: what each gives you out of the box
Vapi’s Logs section brings together calls, chats, sessions, webhooks and API requests, with raw data kept 14, 30 or 180 days on Usage only, Core and Pro. On top sit Boards dashboards, Monitoring alerts by email, Slack or webhook, simulations and a native Langfuse integration.
Pipecat gives you instruments to wire up: TTFB, time to first audio and processing time per service when you set enable_metrics=True, plus OpenTelemetry tracing, and Pipecat Evals, which runs scripted conversations and simulated callers against the same agent you deploy. Pipecat Cloud adds sessions, per-session CPU and memory metrics, agent logs and a Datadog guide, though its logging page says “This section of the documentation is currently work in progress”, and the Datadog guide “may have some sharp edges”. Expect to build dashboards and alerts yourself.
Self-hosting, regions and HIPAA on Vapi vs Pipecat
Pipecat runs anywhere Python runs; Pipecat Enterprise puts agents in your own Kubernetes cluster, where Daily lists the scaling ceiling as “No platform-imposed ceiling” and secret values never leave the cluster. Pipecat Cloud’s Daily-hosted regions, us-west, us-east, eu-central (Frankfurt) and ap-south (Mumbai), are open to every organisation. Vapi has no on-premise option, and its EU region is available only with Premier.
For HIPAA, Vapi lists a $2,000 a month add-on, charged per organisation, that comes with a standard Data Processing Agreement and Business Associate Agreement. Pipecat Cloud’s HIPAA page says Daily can sign one BAA covering Pipecat Cloud, Daily WebRTC and Daily transcription, and that with HIPAA controls on “No application data is logged”. The Pipecat Cloud pricing page lists no HIPAA price; Daily’s Video SDK page lists “Healthcare (HIPAA + BAA): $500/month”, which we could not confirm applies to Pipecat Cloud, as at 18 November 2026.
The sentence on each vendor’s pages that cuts against this page
For Pipecat Cloud it is capacity. The pricing page says “Pipecat Cloud supports unlimited concurrency”, but its scaling guide says “Each deployment made to a Daily-hosted region has a maximum allowed pool size of 50”, raised through a request form. The same guide warns that if no instance is ready within about 2 minutes, “the session is dropped”, and that “Around 10 seconds is a best case” for a cold start. A receptionist that must answer the first ring needs warm agents.
For Vapi it is the fee that never falls at list price: $0.05 on every minute, on every package, plus $10 a month for each line above the package default, and a flat no to on-premise. A team running hundreds of thousands of minutes a month pays for orchestration it could own.
Which to choose: platform, done-for-you, Vapi or Pipecat
Methodology: Zian AI is listed first as the platform pick by this site’s editorial policy, from its published capabilities only. Vapi and Pipecat facts come from each vendor’s own pricing pages, documentation and GitHub repository, read on 18 November 2026, under the same evidence rule for both. No scores are assigned. Disclosure: zian.ai publishes this comparison and Zian appears in it.
| # | Option | Best for | What you run |
|---|---|---|---|
| 1 | Zian AI | Teams that want AI reception and outbound agents without building a voice stack | Nothing: inbound calls 24/7, outbound calls, SMS, email, WhatsApp, Instagram DMs and Facebook Messenger on one platform |
| 2 | LeadsNow.ai | Done-for-you appointment setting, pay-per-result | Nothing: a managed service that runs on Zian |
| 3 | Vapi | Developers (with a no-code editor for operators) who want configuration, webhooks and packaged outbound campaigns, without owning media infrastructure | Your webhook server; Vapi runs the calls |
| 4 | Pipecat + Pipecat Cloud | Python teams that want the agent as code, open turn models, self-hosting, or volumes where the Four-Cent Test pays | Your pipeline, on Pipecat Cloud, Pipecat Enterprise or your own servers |
Vapi is wrong for you if you need on-premise deployment, EU processing below Premier, or a list fee that falls with volume. Pipecat is wrong for you if nobody will own a Python audio service in production, if a non-developer must change the agent after launch, or if you want a packaged list dialler without writing one.
Want booked appointments, not a stack to run? Apply For Partnership
Where every figure on this page comes from
| Figure | Who published it | Link | Date read |
|---|---|---|---|
| Vapi hosting fee $0.05/min; models at provider price, no markup; four per-minute components; $5 free credits | Vapi | https://vapi.ai/pricing | 18 November 2026 |
| Vapi packages: Core $29/month; Pro 10% of hosting, $999/month minimum; Premier custom; $10 per extra line; $2,000/month HIPAA per organisation; Premier Agent Strategist and Forward Deployed Engineer Advisor | Vapi | https://vapi.ai/pricing | 18 November 2026 |
| Vapi calculator ranges: Deepgram $0.00953–$0.00994, OpenAI $0.00774–$0.04523, Cartesia $0.00689–$0.01294 per min | Vapi | https://vapi.ai/pricing | 18 November 2026 |
| Vapi transport: Telephony/SIP, WebSockets, Daily WebRTC free; Twilio $0.008 in / $0.014 out, Vonage $0.00814, Telnyx $0.0055 | Vapi (citing providers) | https://vapi.ai/pricing | 18 November 2026 |
| Vapi included lines 4 / 10 / 30; numbers 1 / 5 / 10; retention 14 / 30 / 180 days; SLAs (two-business-day; one-business-day + 99%; one-hour P0 + 99.9%); named support team; EU region Premier only | Vapi | https://docs.vapi.ai/billing/pricing-and-success-packages | 18 November 2026 |
| Vapi Server URL needed for function calling; no on-premise; volume pricing available | Vapi | https://docs.vapi.ai/faq | 18 November 2026 |
| Vapi free numbers US-only, inbound only, no campaigns, no port-out | Vapi | https://docs.vapi.ai/free-telephony | 18 November 2026 |
| Vapi SIP integration guides: Twilio, Telnyx, DIDWW, didlogic, Bitcall, Zadarma, Plivo, Amazon Chime SDK | Vapi | https://docs.vapi.ai/llms.txt | 18 November 2026 |
| Vapi custom “BYO” SIP trunk | Vapi | https://docs.vapi.ai/advanced/sip/sip-trunk | 18 November 2026 |
| Vapi number imports: Twilio, Telnyx, DIDWW | Vapi | https://docs.vapi.ai/phone-calling | 18 November 2026 |
| Vapi Transfer Call tool: blind (default) and warm modes, phone number or SIP URI | Vapi | https://docs.vapi.ai/tools/transfer-call | 18 November 2026 |
| Vapi transcriber estimate includes two audio channels | Vapi | https://docs.vapi.ai/assistants/model-intelligence/understanding-cost | 18 November 2026 |
| Vapi speaking plans: 0.4 s wait, endpointing options, 0–15 s custom timeout, stop settings 0.2 s and 1 s | Vapi | https://docs.vapi.ai/customization/speech-configuration | 18 November 2026 |
| Vapi WebSocket transport: PCM or mu-law | Vapi | https://docs.vapi.ai/calls/websocket-transport | 18 November 2026 |
| Vapi Outbound Campaigns: up to 10,000 contacts, schedule, concurrency | Vapi | https://docs.vapi.ai/outbound-campaigns/overview | 18 November 2026 |
| Vapi Logs, Boards, Monitoring, simulations, Langfuse integration | Vapi | https://docs.vapi.ai/observability/logs/overview | 18 November 2026 |
| Vapi Boards dashboards | Vapi | https://docs.vapi.ai/observability/boards-quickstart | 18 November 2026 |
| Vapi Monitoring alerts by email, Slack or webhook | Vapi | https://docs.vapi.ai/observability/monitoring-quickstart | 18 November 2026 |
| Vapi simulations | Vapi | https://docs.vapi.ai/observability/simulations-overview | 18 November 2026 |
| Vapi native Langfuse integration | Vapi | https://docs.vapi.ai/providers/observability/langfuse | 18 November 2026 |
| Vapi CX: code-level control for engineers, a no-code editor for operators | Vapi | https://vapi.ai/products/vapi-cx | 18 November 2026 |
| Vapi traffic splitting (12 October 2026 release); release notes to 9 November 2026 | Vapi | https://docs.vapi.ai/whats-new | 18 November 2026 |
| Pipecat Cloud agent-1x $0.01 active / $0.0005 reserved; agent-2x $0.02; agent-3x $0.03; unlimited concurrency; Daily WebRTC voice free; Daily PSTN $0.018; Daily SIP $0.003–$0.02; SIP REFER $0.20; transcoding and transfer conditions; Krisp VIVA free to 10,000 min then $0.0015; Enterprise volume pricing and 24/7 support | Daily (Pipecat Cloud) | https://www.daily.co/pricing/pipecat-cloud/ | 18 November 2026 |
| Daily PSTN $0.018–$0.03 per min; phone number $2/month; Healthcare (HIPAA + BAA) $500/month | Daily | https://www.daily.co/pricing/video-sdk/ | 18 November 2026 |
| Pipecat Cloud: pool maximum 50 per deployment; 1,440 reserved minutes a day; metered to the second; ~10 s best-case cold start; session dropped after ~2 min; free auto-scaling buffer | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/fundamentals/scaling | 18 November 2026 |
| Pipecat framework BSD-2, “works with any AI provider and any hosting environment”; built and maintained by Daily | Pipecat (Daily) | https://docs.pipecat.ai/overview/introduction | 18 November 2026 |
| Pipecat Cloud and Pipecat Enterprise: regions in the US, Europe and India; your Kubernetes cluster | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/introduction | 18 November 2026 |
| Pipecat Enterprise: no platform-imposed ceiling; secrets stay in your cluster | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/enterprise/overview | 18 November 2026 |
| Regions us-west, us-east, eu-central, ap-south | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/guides/regions | 18 November 2026 |
| Pipecat telephony connection types and carriers | Pipecat (Daily) | https://docs.pipecat.ai/pipecat/telephony/overview | 18 November 2026 |
| Framework “orchestrates AI services (STT, LLM, TTS, and more) into real-time pipelines”; 15 frame serializers; turn-1-mini, Krisp VIVA Turn, Alkove Turn; deploy from GitHub; no campaign entry | Pipecat (Daily) | https://docs.pipecat.ai/llms.txt | 18 November 2026 |
| Dial-out to phone numbers is a paid Daily feature, enabled per domain through a request form; international dial-out enabled per domain through sales | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/guides/telephony/daily-dial-out | 18 November 2026 |
| Smart Turn v3 default stop strategy since v0.0.102; weights bundled | Pipecat (Daily) | https://docs.pipecat.ai/api-reference/server/utilities/turn-detection/smart-turn-overview | 18 November 2026 |
| Pipecat metrics (TTFB, TTFA, processing time) | Pipecat (Daily) | https://docs.pipecat.ai/pipecat/fundamentals/metrics | 18 November 2026 |
| Pipecat OpenTelemetry tracing | Pipecat (Daily) | https://docs.pipecat.ai/api-reference/server/utilities/opentelemetry | 18 November 2026 |
| Pipecat Cloud logging “work in progress”; per-session CPU and memory metrics | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/fundamentals/logging | 18 November 2026 |
| Datadog guide “may have some sharp edges” | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/guides/using-datadog | 18 November 2026 |
| Pipecat Cloud HIPAA: one BAA covering Pipecat Cloud, Daily WebRTC and transcription | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/security/hipaa | 18 November 2026 |
| Forward Deployed Engineering in Daily’s enterprise support | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/support | 18 November 2026 |
| BSD 2-Clause licence, copyright Daily; transport list | Pipecat (GitHub) | https://github.com/pipecat-ai/pipecat | 18 November 2026 |
| pipecat-ai 1.16.0 released 12 November 2026 | Python Package Index | https://pypi.org/project/pipecat-ai/ | 18 November 2026 |
| Pipecat Cloud WebSocket endpoints for Twilio, Telnyx, Plivo, Exotel plus a generic endpoint | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/guides/telephony/overview | 18 November 2026 |
| Pipecat Cloud WhatsApp Business voice calls via the WhatsApp Cloud API and SmallWebRTC | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/guides/whatsapp | 18 November 2026 |
| Daily PSTN cold and warm transfer examples | Pipecat (Daily) | https://docs.pipecat.ai/pipecat/telephony/daily-pstn | 18 November 2026 |
| Smart Turn on Pipecat Cloud: “Free to use: The model is fully open source” | Pipecat (Daily) | https://docs.pipecat.ai/pipecat-cloud/guides/smart-turn | 18 November 2026 |
| User turn start strategies (VAD, transcription, minimum words, wake phrase, Krisp VIVA interruption prediction) | Pipecat (Daily) | https://docs.pipecat.ai/api-reference/server/utilities/turn-management/user-turn-strategies | 18 November 2026 |
| Pipecat Evals: scripted conversations and simulated callers | Pipecat (Daily) | https://docs.pipecat.ai/pipecat/evals/overview | 18 November 2026 |
| US$1,000 per engineer-day; 3 + 2 engineer-days a month; $3,458 a month amortised build | Zian AI (our own estimate) | https://zian.ai/voice-agent-platform-vs-framework-when-to-switch/ | 18 November 2026 |
| Itemised bills, Four-Cent Test, platform gaps, crossovers and telephony scenarios (our arithmetic on the figures above) | Zian AI (this page) | — | 18 November 2026 |
Frequently asked questions
Is Pipecat cheaper than Vapi?
On published platform fees, yes. Our worked bill comes to $363.20–$802.70 a month on Pipecat Cloud against $770.60–$1,210.10 on Vapi at 10,000 minutes, with the same models, list prices and the carrier left out. That leaves out the engineering time to write and run a Pipecat agent, which on our own estimates outweighs the saving below roughly 119,000 minutes a month.
Is Pipecat free?
The framework is. Pipecat’s documentation says it is free to use under the BSD-2 licence and works with any AI provider and any hosting environment. You still pay your model providers and whoever hosts the agent; Pipecat Cloud charges $0.01 a minute for an active agent-1x voice agent and $0.0005 a minute for each warm reserved agent, as at 18 November 2026.
How much does Vapi charge per minute compared with Pipecat Cloud?
Vapi charges a $0.05 per minute hosting fee on every minute, plus models at the provider’s listed price with no markup, as its pricing page states. Pipecat Cloud charges $0.01 per active minute on agent-1x, with models billed by your providers directly, according to Daily’s Pipecat Cloud pricing page. Both were read on 18 November 2026, and both vendors say volume pricing is available.
Does Pipecat Cloud charge for idle agents?
Only for warm ones you ask for. Each reserved agent set with min-agents bills $0.0005 a minute on agent-1x around the clock, which Pipecat’s scaling guide puts at 1,440 reserved minutes a day even with zero active sessions. Set min-agents to 0 and you pay nothing while idle, but the first call waits for a cold start of around 10 seconds at best.
Can Pipecat make phone calls to real numbers?
Yes. Pipecat connects to Twilio, Telnyx, Plivo and Exotel over WebSocket media streams, to Daily’s own PSTN numbers for dial-in and dial-out, and to SIP. On Pipecat Cloud, Daily PSTN is listed at $0.018 a minute, and Twilio minutes are billed by Twilio.
Can I self-host Vapi or Pipecat?
Pipecat, yes: the framework runs on any infrastructure, and Pipecat Enterprise runs Pipecat Cloud agents in a Kubernetes cluster you operate. Vapi, no: its FAQ states that Vapi does not support on-premise deployments.