[Digital Today reporter Chi-gyu Hwang] Chinese companies such as Alibaba, ByteDance and Kuaishou have successively unveiled new AI models, CNBC reported on Friday.

Alibaba unit DAMO Academy unveiled RynnBrain, an AI model that allows robots to perceive their surroundings and identify objects. A demo video showed a robot picking up an orange and placing it in a basket, and taking milk out of a refrigerator.

The company said RynnBrain was designed to strengthen AI time-and-space awareness so it can deliver stable performance even in complex tasks. Adina Yakefu (아디나 야케푸), a researcher at Hugging Face, said, "RynnBrain goes beyond simple reactions. It can remember when and where events occur and continue to carry out tasks."

ByteDance announced Seedance 2.0, an AI model that generates realistic video using text alone. It can also use user-provided videos or images. Billy Boman (빌리 보만), an AI content creator, said, "Until 2023, AI video was short and low-quality, but now you can easily generate realistic video." Some Seedance 2.0 features were suspended after its ability to generate voice based on portraits became embroiled in controversy.

Kuaishou unveiled Kling 3.0, which generates realistic 15-second video and multilingual audio. It directly competes with Seedance 2.0, and Kuaishou shares have risen more than 50 percent recently. Zhipu AI also announced GLM-5, an open-source large language model with strengthened coding capabilities, and MiniMax introduced its M2.5 open-source model with enhanced AI agent functions.