Tavus
Real-time AI humans that see, hear and talk back on video
Overview
Tavus is a developer platform for real-time conversational video: AI agents with a lifelike face and voice that see, hear and respond in live video calls, built on its own Phoenix rendering, Sparrow turn-taking and Raven perception models. It also offers custom Faces trained from video or a photo, scripted video generation by API, and a consumer PALs app.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 14 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- Whitelabeled APIs
- 25 minutes of AI conversational video
- 5 minutes of AI video generation
- 25 stock replicas; 30+ languages
- 3 custom replica trainings a month ($65 each beyond)
- 100 minutes of conversational video ($0.37/min beyond)
- 10 minutes of video generation ($1/min beyond)
- Up to 3 concurrent streams
- 7 custom replica trainings a month ($40 each beyond)
- 1,250 minutes of conversational video ($0.32/min beyond)
- 100 minutes of video generation; 100+ stock replicas
- Conversation recordings; up to 10 concurrent streams
- 100% white-labeled experience
- Custom concurrency limits
- Guaranteed SLAs for speed and compute
- SOC 2 and HIPAA compliance reports
- Unlimited messaging
- Support for 30+ languages
- MCP early access on Plus and Max
Use cases
What you can produce with Tavus
- A real-time, face-to-face AI video agent embedded in your app with React components, an embed element or Daily's SDKs
- A custom Face trained from about two minutes of video or from a single headshot
- Agents grounded in your documents through a knowledge base, with memory across conversations
- Objectives, guardrails and function calling that connect conversations to your own APIs
- Scripted or audio-driven videos of a Face generated through the Generate Video API
- AI interviewer, tutor, SDR or support personas that join Google Meet, Zoom or Teams calls
- A free developer plan for prototyping before paid, usage-based plans
Tavus is the developer pick for live, face-to-face AI video: an end-to-end pipeline for rendering, turn-taking and perception, a knowledge base, memories, guardrails and function calling, and embeds for React, plain HTML or Daily. The free plan is enough to prototype, and prices are published from $59 a month plus usage.
It is an API platform, not an editor, so marketers wanting a click-to-create avatar studio will do better elsewhere. Treat the Griffin Turing-test claim as a vendor study, and budget for per-minute overage once conversations scale.
What we found
Tavus is a San Francisco company that went through Y Combinator's Summer 2021 batch. It started out making personalised sales videos and now calls itself a human computing company. Its core product is the Conversational Video Interface (CVI): real-time, face-to-face AI agents, which Tavus brands PALs, that see, hear and answer on video. Hassaan Raza and Quinn Favret started it in 2020. In November 2025 it announced a $40 million Series B led by CRV, with Scale Venture Partners, Sequoia Capital, Y Combinator, HubSpot Ventures and Flex Capital; Tavus puts its total raised at about $64 million, more than the $58 million earlier versions of this page used.
The model stack moves quickly. Phoenix-4.5, published 10 September 2026, renders the face and now the upper body; Sparrow-2, published 27 August 2026, handles turn-taking; Raven-1 reads expressions and surroundings. Griffin-Lite, a single video-to-video model, is a research preview open only to selected testers. Tavus calls it the first model to pass a real-time video Turing test, because 48% of 54 participants in its own study thought they had spoken to a person; that is a vendor study rather than an outside benchmark. Developers create a Face, the docs' new name for a replica, from about two minutes of video or a single headshot, then add a knowledge base, memories, objectives, guardrails and function calling, bring their own LLM if they like, and embed the result with React components, an embed element or Daily's SDKs.
Pricing is published. The free developer plan includes 25 conversation minutes, 5 generation minutes and 25 stock replicas. Starter is $59 a month plus usage and Growth $397 a month plus usage, with conversation overage at $0.37 or $0.32 a minute. A separate consumer PALs app costs $0, $20 or $50 a month.
The review-site complaints this page used to summarise, about bugs, flat voices and thin documentation, could not be traced to specific reviews and were removed.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
About this listing
This entry was compiled from publicly available data including Tavus's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Tavus unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Tavus directly →
Spotted something out of date? Suggest an update →
Alternatives to Tavus
More in Video & Audio
