
Developer APIs for AI video replicas and real-time conversational video
Tavus is API infrastructure for AI video replicas and real-time conversational agents, built for engineering teams shipping a product rather than sales teams running campaigns.
Best for
Engineering teams embedding AI video replicas or real-time conversational video agents into their own product.
Not great for
Sales teams that want a finished outreach workflow, since there is no sequencer, contact data, delivery or booking layer.

Starting price
$59/mo
Pricing model
Credit packs
Free trial
Free plan instead
Platforms
API, Web
Founded
2020
Headquarters
United States
Tavus, in plain terms
Tavus is a developer platform for AI video. Its models cover two jobs: asynchronous video generation from a replica of a real person, and the Conversational Video Interface, a real-time layer that gives an AI agent a face and a voice for live face-to-face interaction. Everything is exposed through APIs that can be whitelabeled and embedded in your own product, with stock replicas available alongside custom replicas trained on your footage, and 30+ languages listed. This is infrastructure rather than a finished go-to-market application. There is no sequencer, no unified inbox, no contact database and no booking product, so a sales use case means building the workflow around the API yourself or wiring it into tools you already run. Pricing reflects the developer model: a free Basic tier with 25 conversational minutes and stock replicas, then paid tiers that bundle minutes, custom replicas and concurrent streams, with per-minute overage on top and custom enterprise terms above that. A separate consumer product line, PALs, is priced independently. Tavus fits engineering teams building an AI video product or an interactive agent experience. Teams that want personalized video prospecting without writing code will spend most of their time assembling the surrounding stack.
Key features
- Custom AI replicas trained on your footage
- Stock replica library
- Conversational Video Interface for live agents
- Whitelabeled APIs with SDKs and docs
- Per-minute usage pricing with published overage
- 30+ languages
- Concurrent stream limits per tier
- Free developer tier
Pros
- Custom replicas plus a stock replica library, exposed through documented APIs and SDKs
- The Conversational Video Interface gives an AI agent a face and a voice in real time, their clear specialty
- APIs can be whitelabeled and embedded in your own product
- A free Basic tier with 25 CVI minutes lets you evaluate before paying
- 30+ languages listed
- Published per-minute overage rates make usage forecasting straightforward
Cons
- No sequencer, no unified inbox, no contact database and no delivery layer: everything around the video is yours to build (their site)
- Usage is billed on top of the plan, with conversational video overage at $0.37 per minute on Starter and $0.32 on Growth (their pricing page)
- Custom replicas are capped per tier, 3 on Starter and 7 on Growth (their pricing page)
- Growth jumps to $397 per month, a steep step for teams that outgrow the 100 minutes on Starter (their pricing page)
- No official Zapier, n8n or Pabbly app, so no-code automation means community-built connectors or your own code (their docs)
Last round
Series B, $40M, November 2025, led by CRV (reported)
Cost at scale
2,000 conversational video minutes a month is $397 on Growth plus about $240 of overage at $0.32 a minute, roughly $637 a month (published pricing)
Total funding
About $64M raised (reported)
Users claimed
100,000+ developers claimed on their site
| Plan | Price | Limits | Includes |
|---|---|---|---|
| Basic | 0 per month | 25 CVI minutes per month, 25 stock replicas, no custom replicas | 25 CVI minutes, 25 stock replicas, Whitelabeled APIs, 30+ languages |
| Starter | 59 per month | 100 CVI minutes, 3 custom replicas, 3 concurrent streams, $0.37 per extra minute | 100 CVI minutes, 3 custom replicas, 3 concurrent streams, $0.37/min overage |
| Growth | 397 per month | 1,250 CVI minutes, 7 custom replicas, 10 concurrent streams, $0.32 per extra minute | 1,250 CVI minutes, 7 custom replicas, 10 concurrent streams, $0.32/min overage, 100+ stock replicas |
| Enterprise | custom | Unlimited custom replicas, custom concurrency, guaranteed SLAs | Unlimited custom replicas, 100% whitelabeled, custom concurrency, Guaranteed SLAs |
Developer plans from their published pricing page, August 2026: Basic free with 25 CVI minutes and 25 stock replicas, Starter $59/mo with 100 CVI minutes, 3 custom replicas and $0.37/min overage, Growth $397/mo with 1,250 CVI minutes, 7 custom replicas and $0.32/min overage, Enterprise custom. Usage is billed on top of the plan. Check their pricing.
What reviews praise
- Product Hunt 5.0 from 4 reviews
- Realistic avatars with convincing lip sync (Product Hunt)
- Responsive, customer-centric support (Product Hunt)
- Easy onboarding even for non-technical users (Product Hunt)
What reviews flag
- No review-sourced criticism was found; the public review base is too small to characterize
Distilled from public reviews and community threads. Not Weezly's opinion, and not a rating.
API access
API-first: full API access is included on every plan, including the free tier, alongside a no-code portal; concurrency, conversation recordings, white labeling and compliance access are what get gated
Compliance
SOC 2 Type 2, GDPR and HIPAA with BAAs claimed, with HIPAA commercially gated to Enterprise; ISO 27001 is not claimed
Data export
Conversation recordings export to your own Amazon S3, Google Cloud Storage or Azure Blob Storage, with recording and transcript webhooks; a zero data retention option is advertised for Enterprise
Automation platforms
No official Zapier, n8n or Pabbly app; three community-built Make integrations exist; their automation story is developer-first, with an MCP server, a CLI and webhooks
Integrations
- HQ
- San Francisco, California, United States, with teams in London and Munich
- Founded
- 2020 per LinkedIn, 2021 per Y Combinator
- Employees
- 51-200 (LinkedIn)
- Leadership
- Hassaan Raza, Co-founder and CEO; Quinn Favret, Co-founder and COO
- Company page
The full comparison
Weezly vs Tavus: every claim, sourced.

Personalized AI video generation from a template, a CSV or an API call

Video email veteran: record in Gmail or Outlook, send an animated preview

Video sales platform with recording, email and SMS campaigns and a built-in CRM

AI video studio with a large avatar library, cloning and video translation

Video recording, editing and personalization for sales, marketing and support teams

Async screen and camera recording with instant share links, owned by Atlassian
Not directly. Tavus sells APIs for AI video replicas and real-time conversational video agents. Sales teams can build personalized video prospecting on top of it, but the sequencer, contact data, delivery and booking layers are not part of the product.
CVI is Tavus's real-time layer: it gives an AI agent a face and a voice so a user can talk to it live, face to face, inside your application. Plan minutes and concurrent stream limits are set per tier.
From their published pricing in August 2026, developer plans are Basic free with 25 CVI minutes, Starter $59 per month with 100 minutes and 3 custom replicas, Growth $397 per month with 1,250 minutes and 7 custom replicas, and custom Enterprise. Usage above the included minutes is billed per minute, $0.37 on Starter and $0.32 on Growth. Check their pricing page.
Yes. Custom replicas are trained on footage of a real person and used through the API. Stock replicas are also available on every tier, with the count rising on higher plans.
One of these tools is ours.
Weezly is in this directory too: LinkedIn outreach with AI clone videos, a 300M+ leads database and booking, from $59 per month. See if it earns a spot in your stack.