Tavus pricing: what each plan includes and what you really pay
Tavus pricing runs from a free plan with 20 minutes of conversation a month to Starter at $22, Builder at $59, Growth at $397 and Business at $975 a month, with Enterprise on a custom quote. The bill depends mostly on minutes of real-time conversation through its Conversational Video Interface (CVI): each plan includes a monthly allowance, and from Builder up, extra minutes cost $0.35 to $0.26 each.
We checked every number on Tavus's own pricing page and docs on 18 November 2026; the page also repeats an older, shorter price list in places, so we use its main plan table. We make Life Inside, a website chat agent with an optional talking avatar that is built on HeyGen, not Tavus, so we compete with Tavus for some buyers. This page covers the plans, what drives the bill, a worked example, and what a real-time avatar on your website costs all-in. More tools: AI avatar tools compared.
Tavus pricing plans at a glance
| Plan | Price | What you get | Extra minutes |
|---|---|---|---|
| Free | $0 | 20 minutes of conversational video a month, 1 session at a time, conversations up to 5 minutes, stock faces, the API and the no-code platform, 1 team seat | Not available |
| Starter | $22 a month | 60 minutes a month, 1 custom face slot, 1 session at a time, conversations up to 10 minutes, full transcripts, 1 Google Meet, Teams or Zoom configuration, 5 minutes of video generation | Not available: Starter has no pay-as-you-go overage |
| Builder | $59 a month | 175 minutes a month, 3 custom face slots, 25 stock faces, up to 3 sessions at a time, conversations up to 15 minutes, 15 minutes of video generation | $0.35 a minute. Video generation: $1 a minute |
| Growth | $397 a month | 1,300 minutes a month, 7 custom face slots, up to 10 sessions at a time, conversations up to 60 minutes, conversation recordings, 5 team seats, 50 minutes of video generation | $0.31 a minute. Video generation: $0.90 a minute. Recordings: $0.03 a minute |
| Business | $975 a month | 4,000 minutes a month, 15 custom face slots, up to 15 sessions at a time, conversations up to 60 minutes, priority support and a 1:1 onboarding call, 10 team seats, 250 minutes of video generation | $0.26 a minute. Video generation: $0.80 a minute. Recordings: $0.03 a minute |
| Enterprise | Custom quote | Volume discounts, unlimited custom faces, custom concurrency and conversation length, a fully white-labelled experience, SLAs, SOC 2 and HIPAA reports with BAAs, zero data retention, unlimited team seats | Custom, with scaling discounts |
Which Tavus plan fits
- Free or Starter if you are testing: 20 or 60 minutes a month, one conversation at a time, and no way to buy extra minutes.
- Builder if you are piloting with real users: up to 3 conversations at once and extra minutes at $0.35.
- Growth or Business if conversations are part of your product: 10 to 15 at once, hour-long conversations and lower rates per extra minute.
- Enterprise if you need HIPAA, zero data retention, SLAs or more than 15 people talking at the same time.
- None of them: use [Life Inside](/ai-video-agent) if what you want is an avatar or chat agent that answers visitors on your website, built and run for you, rather than an API to build on.
Tavus CVI: what the Conversational Video Interface is
Tavus sells real-time AI agents with a face. It calls the agents PALs and their appearance faces; older docs and endpoints call them personas and replicas. The Conversational Video Interface (CVI) is the pipeline behind them: speech recognition, a language model, a voice and a generated face, streamed to the user over WebRTC at up to 1080p.
Three of its own models do the face-to-face part: Phoenix-4.5 renders the face, Raven-2 lets the agent see and respond to visual cues from the camera, and Sparrow-2 handles turn-taking. Every plan includes the API and the no-code platform, and Tavus offers a knowledge base built from documents or websites. Tavus speaks 42 languages with its default voice engine, and you can use its language model or bring your own.
To put a PAL on a website you add its embed element and script to the page, or use a floating widget. PALs can also join Google Meet, Zoom and Microsoft Teams calls.
Tavus API pricing: what the bill depends on
- Conversation minutes. Each plan includes a monthly allowance. Starter stops at 60 minutes; Builder, Growth and Business bill extra minutes at $0.35, $0.31 and $0.26.
- A 30-second minimum per conversation. A visitor who leaves after 10 seconds is billed for 30. Extra minutes are rounded to the nearest 6 seconds.
- Concurrency. Free and Starter allow one conversation at a time, Builder 3, Growth 10 and Business 15. More needs Enterprise.
- Conversation length. Conversations end at 5 minutes on Free, 10 on Starter and 15 on Builder; Growth and Business allow 60.
- Custom faces. A custom face is trained from about one minute of video and includes a voice model. Plans include 1, 3, 7 or 15 face slots.
- Video generation is separate. Async video has its own allowance (5 to 250 minutes a month on paid plans) and its own rate, from $1 down to $0.80 a minute.
- Recordings. Video recordings of conversations cost $0.03 a minute on Growth and Business. Transcripts are included.
How to estimate your monthly Tavus cost
Start with conversations a month times their average length. Say an avatar on a product page holds 300 conversations a month of 3 minutes each: 900 minutes.
- Builder: $59 includes 175 minutes. The other 725 minutes at $0.35 add $253.75, so about $313 a month, as long as no more than 3 people talk at once.
- Growth: $397 covers up to 1,300 minutes, so $397 a month, with room for 10 conversations at once.
- Where the plans cross: Builder is cheaper below about 1,140 minutes a month, Growth up to about 3,160 minutes, and Business above that.
- Short conversations cost more than they look. 1,000 conversations of 20 seconds are billed as 500 minutes, not 333, because of the 30-second minimum.
What a Tavus avatar on your website costs all-in
The minutes are only one line. A real-time avatar that answers your visitors also needs a persona with objectives and guardrails, a knowledge base kept up to date, the embed and page design, a language model if you bring your own (paid to that provider), and a way to see what people asked and which answers were missing. Most of all it needs someone, usually a developer, to build it and keep it right.
Life Inside packages that as a managed service. The AI Chat Agent answers in text first, from content you approve, and shows interactive video of real people. The talking avatar, the AI Video Agent, is an add-on built on HeyGen's real-time avatar technology. Life Inside builds the knowledge base, you add one JavaScript snippet, and most agents go live within a day. AgentLoop shows what visitors asked, where they dropped off and what the agent could not answer. Plans are on our pricing page.
When Tavus is the better choice, and when Life Inside is not
Choose Tavus if the conversation is your product: interviews, patient intake, tutoring, sales practice or support inside your own app. Its perception model, meeting integrations and HIPAA option on Enterprise are things Life Inside does not offer. If you have developers and want full control, Tavus's API is built for that.
Life Inside is not the right choice if you need a developer API, want to generate videos, need an avatar inside your own app or in video meetings, or want live video calls between people. It fits when the job is a website that answers visitors. Compare the alternatives on our Tavus alternatives page, see Tavus vs HeyGen, or book a demo.
Frequently asked questions
Sources
- Tavus pricing (plans, minutes, overage, compare table) (Checked 2026-11-18)
- Tavus: PALs, Developer API, PAL Maker and Enterprise (Checked 2026-11-18)
- Tavus docs: introduction (PALs, faces, legacy names) (Checked 2026-11-18)
- Tavus docs: language support (Checked 2026-11-18)
- Tavus docs: embed a PAL on a website (Checked 2026-11-18)
Updated 2026-11-18