Tencent Unveils 13-Billion Parameter Video Generation Model, Open-Sourcing its Capabilities
Tencent’s HunYuan Large Model to LaunchVideo Generation on December 3rd
Tencent is set to significantly expand the capabilities of its HunYuan large language model (LLM) with the launchof a powerful video generation feature on December 3rd, according to Zhang Jun, Tencent’s Public Relations Director. This announcement, made late on November28th, signals a major step forward in Tencent’s AI strategy and promises to democratize access to advanced video creation tools.
The new video generation capabilities will be initially accessible through two channels. Individual users can apply fortrial access via the Tencent Yuanbao app, a recently updated platform featuring a dedicated AI applications section. This section already includes AI search, AI reading, creative drawing, inspiration galleries, and AI photo enhancement tools, highlighting Tencent’s commitmentto integrating AI into its ecosystem. Furthermore, the AI search function now incorporates resources from WeChat Video, QQ Music, and other platforms, significantly broadening its scope.
For enterprise clients, access will be provided through Tencent Cloud, with the API concurrently opening for internal testing applications. This dual-pronged approach aimsto cater to both individual creators and businesses seeking to leverage the power of AI-driven video generation.
Crucially, Tencent plans to open-source the underlying video generation model, boasting a substantial 13 billion parameters. This move underscores Tencent’s commitment to fostering innovation within the AI community and accelerating the developmentof advanced video generation technology. The open-sourcing also allows for community scrutiny and potential improvements to the model.
This development follows the mid-November release of Yuanbao 2.0, which introduced a revamped interface, improved historical asset management, and the aforementioned AI application section. The improved user experienceis a clear indication of Tencent’s focus on providing a seamless and intuitive user interface for its AI tools.
The announcement also builds upon the September launch of Tencent HunYuan Turbo, a next-generation large model utilizing a Mixture-of-Experts (MoE) architecture. This model boasts a 100% improvement in inference efficiency and a 50% reduction in inference costs compared to its predecessor, HunYuan Pro. The pricing for HunYuan Turbo is also significantly reduced, costing 0.05 yuan per thousand tokens for output and 0.015 yuan per thousand tokens for input.
The launch of HunYuan’s video generation capabilities marks a significant milestone for Tencent and the broader AI landscape. The combination of readily available access, open-sourcing, and cost-effective pricing positions Tencent to become a major player in the rapidly evolving field of AI-powered video creation. The impact on contentcreation, marketing, and various other industries is expected to be substantial.
Conclusion:
Tencent’s decision to launch and open-source its 13-billion parameter video generation model represents a bold step towards democratizing access to advanced AI technology. This move will likely accelerate innovation in the field ofvideo generation, fostering creativity and efficiency across diverse sectors. The future implications for content creation, marketing, and other industries remain to be seen, but the potential for transformative impact is undeniable. Further observation of the model’s performance and community contributions following its open-sourcing will be crucial in assessing its long-term influence.
References:
- IT之家. (2024, November 28). 张军：腾讯混元大模型 12 月 3 日将上线视频生成能力. [Online]. Available: [Insert IT Home article URL here] (Note: Replace bracketed information with actual URL)
Views: 0