Contents
Fast answer
LiveKit is real-time infrastructure at $0.01 an agent minute, and the alternatives are all further up the stack. Vapi is the nearest neighbour at $0.05 a minute for orchestration with your own model, speech and telephony providers, which is the same philosophy with more of the voice-agent plumbing already assembled. Retell AI at $0.07 to $0.31 a minute gives you a managed pipeline plus a visual builder and publishes every component rate. Bland AI at $0.14 a minute with no monthly fee bundles the model, transcription and premium voices into one number. Vocode is the other open-source option if what you liked about LiveKit was owning the code. ElevenLabs Agents at an effective $0.08 a minute is the choice when speech quality is the differentiator. The question is not price, it is how much of the stack you want to build.
Facts last verified 2026-08
Quick facts
- Why people leave
- It is infrastructure, not a product; there is no dashboard where you go live
- Vapi
- $0.05/min orchestration, providers at cost or $0 with your own keys
- Retell AI
- $0.07 to $0.31/min itemised, plus a visual builder
- Bland AI
- $0.14/min on Start, model and premium voices bundled
- ElevenLabs Agents
- $0.08/min effective on every tier
- LiveKit for reference
- $0.01 per agent minute; Build free with 1,000 minutes
- What LiveKit keeps
- Open-source framework and media server you can self-host
On this page
What this page covers
A comparison visitor should understand the tradeoff, the best-fit scenario, and the next diagnostic tool to confirm the choice.
- 01Why people leave
- 02Vapi
- 03Retell AI
- 04Bland AI
- 05Vocode
- 06ElevenLabs Agents
Why does this matter now?
LiveKit's rate is the lowest in this comparison by a wide margin and its total cost is frequently the highest, because a cent a minute buys orchestration and nothing else. The model, transcription, speech synthesis, telephony and, most importantly, the engineering are all still yours. Teams that leave almost never leave over the bill. They leave because they wanted a working phone agent and got a platform to build one on.
Internal path
Where to go next from this page
These links are part of the A8gent learning and conversion path. Use them to move from concept, to diagnosis, to workflow build, to course.
What you should be able to do after this
- Decide how much of the stack you want to own
- Trade a lower rate for less assembly
- Keep control where compliance requires it
- Know when LiveKit is genuinely the right layer
How do you do it, step by step?
1. Vapi, when you want to keep control but skip the plumbing
Vapi charges $0.05 a minute for orchestration with the model, speech-to-text and speech synthesis billed by your own providers, or at nothing to Vapi with your own API keys, plus $0.005 per SMS or chat message. It keeps LiveKit's bring-your-own-provider philosophy while handling far more of the voice-specific assembly, which is usually the gap that stalls LiveKit projects.
2. Retell AI, when you want a pipeline and a builder
Retell publishes voice infrastructure at $0.055 a minute, speech at $0.015, telephony at $0.015 for US Twilio, and a model from $0.003 to $0.16, for an all-in $0.07 to $0.31. The first 20 concurrent calls are free. It is a managed pipeline plus a visual builder, so a technically confident operator gets a working agent live without an engineer owning it full time.
3. Bland AI, when you want to stop assembling entirely
Bland's Start plan has no monthly fee at $0.14 a minute with the model, transcription and premium voices bundled and no token charges to reconcile, plus $0.05 a transfer minute and an inbound number. It is the opposite end of the spectrum from LiveKit, and for a business without an engineer that is a feature rather than a compromise.
4. Vocode, when owning the code was the point
If the reason you chose LiveKit was that the framework and media server are open source, Vocode is the other open-source voice agent project to evaluate. Compare them on how much of the voice pipeline each gives you before you write anything, because that is where the engineering weeks actually go, not on the transport layer.
5. ElevenLabs Agents, when speech quality is the requirement
Every ElevenLabs Agents tier works out at $0.08 a minute, from Starter at $6 for 75 minutes to Business at $990 for 12,375, with the model and telephony billed separately and burst at $0.16 when you exceed concurrency. Choose it when a distinctive brand voice is doing commercial work that a default voice would not.
The trigger is that you wanted a product and got a platform
LiveKit's rate is the lowest in voice AI and almost nobody leaves over cost. The trigger is that there is no dashboard where you configure an agent and go live. LiveKit provides real-time media transport and session handling and expects you to supply everything else, including the engineer. That is honest positioning rather than a gap, and it is the reason evaluations stall: a team arrives expecting a voice agent product and finds infrastructure to build one on.
The rate reinforces the misunderstanding rather than correcting it. $0.01 an agent session minute looks like it undercuts every bundled vendor by an order of magnitude, and it does, because it covers a fraction of what they cover. The language model, speech-to-text, text-to-speech and telephony are all separate suppliers you contract with and pay directly. Anyone comparing $0.01 against a bundled $0.14 and concluding LiveKit is fourteen times cheaper has compared two different products.
The second trigger is the meters around the edges, which are easy to miss when modelling. Third-party SIP is $0.004 a minute on Build and Ship and $0.003 on Scale. Downstream data is $0.12 a gigabyte on Ship and $0.10 on Scale. Extra phone numbers are $1.00 a month each beyond the free local number. Bandwidth in particular has no intuitive relationship to anything on your business dashboard and routinely appears larger than expected on a first real month.
The third trigger is the inference credits being read as an inference budget. Build includes $2.50, Ship $5 and Scale $50. Those cover early testing and will not cover a production month of real model usage, after which inference bills per model at provider rates. A forecast built on the assumption that credits cover inference is wrong by an order of magnitude, and that discovery is a common reason a pilot suddenly needs re-approval.
What each alternative fixes
Vapi fixes the assembly gap while keeping the philosophy. It charges $0.05 a minute for orchestration and passes the model, transcription and speech through at provider cost, or at nothing to Vapi with your own keys, plus $0.005 per SMS or chat message. You keep bring-your-own-provider control and skip building turn taking, tool calls and call control yourself. Compliance is priced openly at $2,000 a month for HIPAA and $1,000 for Zero Data Retention, which is a negotiation you can budget rather than enter blind.
Retell AI fixes the need for a full-time engineer. Its all-in band is $0.07 to $0.31 a minute, built from voice infrastructure at $0.055, speech at $0.015 or $0.040 with ElevenLabs voices, US Twilio telephony at $0.015 and a model from $0.003 to $0.16, with the first 20 concurrent calls free. It pairs a managed pipeline with a visual builder, so a technically confident operator gets a working agent live without engineering owning it permanently.
Bland AI fixes assembly entirely by removing the choice. Start has no monthly fee at $0.14 a minute with the model, transcription and premium voices bundled and no token charges, plus $0.05 a transfer minute, 10 concurrent calls, 100 calls a day and an included inbound number. It is the opposite end of the spectrum from LiveKit, and for a business without an engineer that is a feature rather than a compromise.
Vocode keeps the open-source route if that was the point, and ElevenLabs Agents covers the case where speech quality is the actual requirement at an effective $0.08 a minute across every tier with the model and telephony separate. If you are comparing Vocode and LiveKit, compare them on how much of the voice pipeline each gives you before you write anything, because that is where the engineering weeks go, not on the transport layer.
What only LiveKit gives you, and the staffing test
Self-hosting is the capability you cannot get from Vapi, Retell, Bland or ElevenLabs, all of which are managed services. The Agents framework and the media server are both open source and can run on your own infrastructure with nothing paid to LiveKit at all. If self-hosting is a regulatory or data-location requirement rather than a cost preference, that constraint should filter your shortlist before price does, and it leaves you with LiveKit or Vocode.
Be realistic about what self-hosting involves, though, because the word makes it sound like a deployment decision rather than an operational commitment. Running real-time media at scale means capacity planning, network tuning, and being on call for something customers hear in real time. It is a serious undertaking rather than a weekend project. Choose it to satisfy a requirement, never to avoid a $50 platform fee.
The other thing you give up is the compliance flexibility that comes with owning the stack. LiveKit has no packaged compliance offering the way bundled vendors sell HIPAA as a monthly add-on; you build to your own requirements on infrastructure you control. That is more flexible and less immediately budgetable, and which of those matters more depends entirely on whether you have the people to exercise the flexibility.
Which brings it to a staffing question rather than a pricing one, and that is the clean test. Is there an engineer who wants to own the voice stack, who will still be here in six months, and whose time is not already committed elsewhere? If yes, LiveKit is excellent infrastructure at an honest price. If no, a bundled platform costs more per minute and vastly less in total, because the alternative is a half-finished integration that nobody owns and everybody blames.
At a glance
LiveKit Agents alternatives with published pricing, read from each vendor's own pricing page on 20 August 2026
| Alternative | Published price | Reason to switch to this one |
|---|---|---|
| Vapi | $0.05/min orchestration; $0.005 per SMS or chat message; HIPAA $2,000/mo, ZDR $1,000/mo | Same bring-your-own-provider philosophy with the voice plumbing assembled |
| Retell AI | $0.07-$0.31/min itemised; 20 concurrent calls free | Managed pipeline plus a visual builder, so no engineer owns it full time |
| Bland AI | Start $0/mo + $0.14/min, model and premium voices bundled | Stop assembling entirely; a non-developer can get an agent live |
| Vocode | See the platform page | The other open-source route if owning the code was the point |
| ElevenLabs Agents | $0.08/min effective; Pro $99 for 1,238 minutes | Speech quality as the differentiator, with a managed platform around it |
| LiveKit Agents (for reference) | Build $0/1,000 min; Ship $50/5,000; Scale $500/50,000; overage $0.01/min | Model, transcription, speech and telephony are all still yours to buy |
What mistakes should you avoid?
- Comparing $0.01 an agent minute against all-in rates from bundled vendors
- Treating the $2.50, $5 or $50 inference credits as an inference budget
- Starting on LiveKit without an engineer who will still own it in six months
- Self-hosting a real-time media server to save a platform fee rather than to meet a requirement
- Moving to a bundled vendor and then being surprised you cannot swap the model
- Forgetting to export transcripts before switching, since they are your test set for the next platform
FAQ
- What is the best LiveKit Agents alternative?
- Vapi, for most teams. At $0.05 a minute for orchestration with your own providers it keeps the same control-first philosophy while handling much more of the voice pipeline. Retell AI at $0.07 to $0.31 a minute is the pick if you want a managed pipeline and a visual builder instead of an SDK.
- Is anything cheaper than LiveKit?
- Not on the agent minute. LiveKit's $0.01 is the lowest published rate here, and it covers the least: no model, no transcription, no speech synthesis, no telephony. Once you add those, the all-in cost frequently lands above platforms with higher headline rates, and well above them once engineering time is priced.
- Which alternative needs the least engineering?
- Bland AI. Its Start plan has no monthly fee at $0.14 a minute with the model, transcription and premium voices bundled, so a non-developer can get a working agent live. Retell AI is the middle ground, with a visual builder that gets a technical operator a long way before the API becomes necessary.
- Can I keep self-hosting if I switch?
- Only by choosing another open-source project such as Vocode. Vapi, Retell, Bland and ElevenLabs Agents are all managed services. If self-hosting was a compliance or data-location requirement rather than a preference, that constraint should filter your shortlist before price does.
- When is LiveKit the right choice?
- When you have an engineer who wants to own the real-time layer, when your requirements are unusual enough that a packaged agent platform gets in the way, or when data location and self-hosting are contractual. It is excellent infrastructure. It is a poor fit for a business that wanted its phone answered.
Sources & further reading
Done for you
Want this built for your business?
Tell us the workflow. We scope, build, and ship the agent with guardrails and the numbers to prove it worked. The scoping call is free.
Free scoping call. You own the code.
Was this page helpful?

