According to media reports, at the main summit of the 2025 Tencent Global Ecosystem Conference, Tencent announced the latest progress in a number of AI technologies and products, and stated that it will fully open its own AI implementation capabilities and advantageous scenarios through Tencent Cloud, helping to accelerate the empowerment of "easy-to-use AI" across various industries.
Regarding the widely concerned computing power challenge in the industry, Qiu Yuepeng, Vice President of Tencent Group and President of Tencent Cloud, stated that Tencent has completed full adaptation to mainstream domestic chips and will continue to actively participate in and contribute to the open-source community.
In addition, Tencent Cloud has long been committed to the full-stack optimization strategy of software-hardware collaboration. By strengthening the software capabilities of heterogeneous computing platforms and integrating diverse chip resources, it provides cost-effective AI computing power services to support the intelligent transformation of industries.
Tencent's self-developed "Tencent Hunyuan Large Model" continues to iterate. Through data enhancement and synthesis technologies, it has improved the quality and diversity of data; and by adopting more effective post-training and in-test expansion methods, it has enhanced the basic model capabilities of Tencent Hunyuan.
Tencent Hunyuan has also launched four open-source small models, including 0.5B, 1.8B, 4B, and 7B versions, which can run on mobile phones, greatly lowering the threshold for model implementation. In terms of multimodality, Tencent Hunyuan has continuously released a number of models this year, and also launched the industry's first interactive world generation model - "Hunyuan 3D World Model 1.0".
(Reprinted from https://news.eccn.com/）