Tavus launches Griffin: 48% of people mistook this video AI for a human in live calls
rohanpaul_ai · x · 2026-10-02
Tavus introduced Griffin, billed as the first Human Interaction Model (HIM), unifying real-time perception, conversation and expressive video generation into one video-to-video system.
Key points:
- In a live study, 48% of participants believed Griffin was a real person after a one-minute video call — the first model to pass a real-time video Turing test
- Ranked #1 on NVIDIA's independent face-to-face AI test, 37% ahead of the next best AI at reacting in the moment
- It attends to expressions and pauses, not just words, listening while it talks
- Built by CEO Hassaan Raza and research head Ioannis Patras, with a research breakdown and live demo available
Related event: Tavus Launches Griffin, an AI Video Model 48% of Callers Mistake for Human(6 posts)→
More from Multimodal
- VAmoS Pro voice-agent benchmark: Grok leads tasks, GPT-Live fastest, Gemini most noise-robust — davlanade · 2026-10-02
- Tavus demos fully realtime-generated AI agent and a vision of delegating identity — Kyrannio · 2026-10-02
- Boiler Room sets turn into agentic coding music hackathons as dev live-codes Temu Ryoji Ikeda — teropa · 2026-10-02
- fal launches Recast: one-click video character replacement keeping motion and audio — OdinLovis · 2026-10-02
- Fixing Krea2's blotchy noise: an algorithmic mid-frequency removal workflow with exact settings — terrariyum · 2026-10-02
- Magnific shares six Ideogram 4.5 posters with full prompt iteration threads — charis_ai · 2026-10-02