Developers building lifelike video agents for sales or support get a free tier, with costs climbing by the minute.
Kenning's take
What it does
Phoenix-4 is a real-time video model from Tavus that animates a digital human's face, head and shoulders during a live conversation. It generates emotional expressions, active listening cues such as nods and micro-expressions, and head movement from the conversation context, at 1080p and 40 frames a second, listening and responding at the same time. Developers use it through the Tavus API to build video agents for sales, interviews, healthcare and training; Tavus has since listed Phoenix-4.5 as its newer version.
- Real-time full-face rendering at 1080p and 40fps
- Controllable emotional states
- Active listening behaviour while the user speaks
- Full-duplex, listening and responding at once
- Identity-preserving custom Replicas
- Available through the Tavus developer API
Who it suits. Developers and companies building AI video agents that talk face to face with customers.
Key facts
- Category
- AI models
- Part of
- Tavus
- Made by
- Tavus
- Released
- 18 February 2026
- How to get it
- through a Tavus developer account and API; free tier, then paid plans
- Works on
- API, Web
- In the UK
- Yes
- Free plan
- Basic developer plan free
- API price
- conversational video streamed at $0.37 per minute on Starter and $0.32 per minute on Growth
Pricing
| Plan | Price | Billed |
|---|---|---|
| Starter | $59 | month plus usage |
| Growth | $397 | month plus usage |
| Enterprise | Custom pricing | custom |
FAQ
When was Tavus Phoenix-4 released?
Tavus announced Phoenix-4 on 18 February 2026.
How do developers use Phoenix-4?
Through the Tavus developer API, which has a free tier and paid plans from $59 a month plus per-minute usage.
Is there a newer version than Phoenix-4?
Yes. Tavus's site now lists Phoenix-4.5 among its models.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.
Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.
Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.