Desktop Mobile
Artificial Intelligence (AI)

Tencent Debuts World’s Largest Text-to-Video AI

Ken Metral

Ken Metral

— 2 min read
Tencent Debuts World’s Largest Text-to-Video AI
Credit: HunyuanVideo/Tencent Holdings Ltd.

Tencent Holdings, a prominent player in video gaming and social media, has entered the competitive field of AI-driven video generation with its latest innovation, HunyuanVideo. Powered by Tencent’s Hunyuan AI foundation model, this text-to-video generator is now offered for free to both businesses and individual users, the company revealed this week.

A World-First in Video Generation

HunyuanVideo, boasting over 13 billion parameters, is the world’s largest open-source video generation model, according to Tencent. The tool is designed to create hyperrealistic videos with lifelike camera angles and reflections. Demonstrations include a surfer spinning on a wave and hot air balloons rising as campfires glow below, showcasing its ability to render visually compelling scenes.

Features and Capabilities

  • Video Length and Quality: Currently, HunyuanVideo generates videos up to five seconds long in 720p resolution.
  • Multilingual Support: Users can write prompts in English or Chinese.
  • Customization Options: Advanced settings allow adjustments in video style, quality, camera movements, and shot angles.
  • Versatility: Examples include a video of a Chinese woman cooking and an animated cat in a cyberpunk world. However, some prompts may not translate perfectly into video.

HunyuanVideo is accessible via Yuanbao, Tencent’s AI chatbot app, making it user-friendly for various creative needs.

Competition in AI Video Generation

Tencent’s move comes amid increasing competition among Chinese and global tech giants in the AI video generation space.

  • Kuaishou Technology launched its Kling AI Model in June.
  • Alibaba Group is developing Tora, based on OpenAI’s OpenSora model.
  • Tencent’s HunyuanVideo is built on the same Diffusion Transformer architecture as OpenAI’s Sora, introduced in February.

A History of Innovation

HunyuanVideo is the latest development in Tencent’s AI advancements:

  • March: Collaboration with universities to launch an image-to-video tool called Follow-Your-Click.
  • May: Added text-to-image capabilities, enhancing Hunyuan’s performance by 20%.
  • September: Released the foundational Hunyuan model, setting the stage for continuous upgrades.

Implications and Future Prospects

As Tencent pushes the boundaries of AI, HunyuanVideo positions the company as a key player in the global race for advanced video generation. Its free availability aims to democratize access to cutting-edge tools, encouraging widespread adoption among creators and enterprises alike.

With rivals like Kuaishou, Alibaba, and OpenAI in the mix, the competition is heating up, promising even more innovation in the AI video generation arena. Tencent’s commitment to expanding its Hunyuan ecosystem underscores its vision of becoming a leader in AI-powered creativity.

Read more