Only matters if you build video agents on Tavus, where it runs by default; the developer plan is free, but Growth costs $397 a month.
Kenning's take
What it does
Raven-1 is a perception model from Tavus that combines audio and video of a person in real time to judge tone, expression, hesitation and intent, rather than relying on a transcript alone. It replaces Raven-0, which handled visual context only, and keeps its understanding no more than 300 milliseconds out of date. It runs automatically in every Tavus conversation, including apps developers build with the Tavus API.
- Fuses audio and visual signals in one model
- Reads tone, expression, hesitation and pauses
- Understanding kept within 300ms of live
- On by default in all Tavus conversations
- Works alongside Tavus's Phoenix and Sparrow models
Who it suits. Developers building real-time AI video agents with Tavus, for uses such as sales, interviews and healthcare.
Key facts
- Category
- AI models
- Made by
- Tavus
- Based in
- United States
- Released
- 10 February 2026
- How to get it
- built into Tavus conversations, through the Tavus developer API and plans
- Works on
- API (Tavus), Web
- Free plan
- Basic developer plan is free
- API price
- extra conversational video minutes $0.37 a minute on Starter, $0.32 on Growth
Pricing
| Plan | Price | Billed |
|---|---|---|
| Starter | $59 | month plus pay as you go |
| Growth | $397 | month plus pay as you go |
| Enterprise | Custom pricing | custom |
FAQ
When was Raven-1 released?
Tavus announced Raven-1 on 10 February 2026.
How do I use Raven-1?
It runs automatically in Tavus conversations, including those built with the Tavus API.
What changed from Raven-0?
Raven-0 interpreted visual context. Raven-1 combines audio and video together to read tone and expression.
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.