Chinese big tech firms from Tencent to Kuaishou are racing to push generative AI models

KwaiYii LLM from Kuaishou, a top Chinese short video platform operator, has “significantly outperformed GPT-3.5 in overall performance … and is approaching GPT-4 levels in some metrics,” company founder and CEO Ou Cheng Yixiao said. A subsequent earnings call on Wednesday.

01:45

A Chinese AI-generated cartoon series airs on state television.

A Chinese AI-generated cartoon series airs on state television.

GPT-4 is OpenAI’s most advanced LLM publicly accessible.

Cheng said Kwaishu is confident KwaiYii can match GPT-4 within the next six months. The company also said its KeTu text-to-image model outperformed Midjourney V5, an AI image-generating service launched a year ago by San Francisco-based Midjourney.

Meanwhile, Tencent said its Haniwan LLM is now among the best in the world.

Social media and video gaming are expanding the ability of its AI models to create images and videos based on text prompts, the company’s president Martin Lau Chiping said in a conference call with media on Wednesday. This year is important for us.”

The Shenzhen-based firm recently introduced a picture-to-video model through tie-ups with academic partners including the Hong Kong University of Science and Technology and Beijing’s Tsinghua University.

Called Follow-Your-Click, the tool is able to animate any part of an image with a simple text prompt and a mouse click.

Both Tencent and Kuaishou are exploring ways to apply the latest AI technology to their businesses.

Advertisements for Kwaishu Technology at a subway station in Beijing, China. Photo: Bloomberg

Tencent’s Lau said the Hunyuan model has been integrated into some of the firm’s enterprise products. This model is being used by employees for day-to-day tasks such as coding and content creation, as well as optimizing ads to drive clicks.

KwaiYii, a major rival of ByteDance’s doyen, has used KwaiYii to improve automated customer service across several of the company’s businesses, including its online talent and property sales platforms, according to CEO Cheng.

He also said the company is looking to use AI to develop creative models that reduce the barrier to content creation for users, improve video quality and increase production efficiency. Can work as tools.

“Sura’s faster-than-expected launch … opens up enormous opportunities for the short video ecosystem,” Cheng said.

Leave a Comment