AI startup Tavus has developed a model called Griffin that convinced 48% of participants (26 out of 54) during one-minute video calls that they were speaking with a real human, compared to just 2.4% for its previous version. Griffin-Lite ranks first on NVIDIA’s VideoFDB benchmark with a score of 3.83 out of 5 for generation, outperforming the next-best system at 2.80. Tavus raised 40 million dollars in a Series B funding round in November 2025, led by CRV. Average audio-to-video delay is 0.43 seconds on NVIDIA H100 chips, half that of the next fastest method. The system remains in limited testing with select trusted users while safety measures are being developed.
Source: Read the original article
Disclaimer: this content is for information purposes only and is not financial advice. Cryptocurrencies are highly volatile: you may lose all of your capital. Always do your own research. Legal notice
