Tavus releases Griffin, a real-time face-to-face video model
What happened
On October 2, The Decoder reported that Tavus released Griffin, a real-time face-to-face conversation model that handles speech, expression, tone, gesture and pauses while receiving and generating video. In the company's study, 48% of participants thought Griffin was human after a one-minute video call, against a previous high of 2%. Tavus also cites an independent Nvidia test where Griffin scored 3.83 on humanness, humans 3.92 and the best prior AI model 2.80.
Written by AI from the coverage · updated 1 hour ago
Coverage
Follow the reports to see the story from different sides.
- The DecoderNearly half of test subjects mistook Tavus' AI video avatar for a real person on a one-minute call
Tavus 推出实时面对面对话模型 Griffin，公司研究中48%的参与者在一分钟视频通话后认为它是真人，此前系统最高为2%。Griffin 在接收和生成视频时处理语音、表情、语调、手势与停顿；在 Tavus 所称的独立 Nvidia 测试中，其拟人程度得分为3.83，真人为3.92，此前最佳 AI 模型为2.80。
Heat over time
Not enough continuous observations to draw a trend yet.