ZhiPu AI Unveils New Generation of Foundation Models, Qingyan AppPioneers Video Calls for C-End Users in China
Beijing,August 29, 2024 – ZhiPu AI, a leading Chinese artificial intelligence company, has announced the release of its new generation offoundation models at the KDD 2024 conference. The company claims these models have achieved international first-tier standards in their respective fields.
The new models include:
- GLM-4-Plus: A language model with significant improvements in language understanding, instruction following, and long-text processing. It is said to be on par with GPT-4o, a leading model in the field. ZhiPu achieved this by implementing a more precise strategy of mixing long and short text data during the model’s training.
- CogView-3-Plus: A text-to-imagemodel that boasts performance comparable to top-tier models like MJ-V6 and FLUX. ZhiPu achieved this by using a Transformer architecture instead of the traditional UNet architecture to train the diffusion model, and by conducting in-depth research on noise planning for diffusion models.
- GLM-4V-Plus: An image/video understanding model with exceptional image comprehension capabilities and the ability to understand videos based on time-based data acquisition. This model will be available on the open platform (bigmodel.cn) and will be the first general-purpose video understanding model API in China.
- CogVideoX:A video generation model that has further enhanced its performance with the release of its 5B version, following the open-source release of its 2B version.
ZhiPu is also making its models readily accessible to developers and researchers by offering free access to the GLM-4-Flash API on its MaaSplatform. This API provides powerful and comprehensive services for inference.
Beyond the release of its new models, ZhiPu has also announced the launch of video calling functionality on its Qingyan App. This marks the first time a video call service has been made available to C-end users in China.
The GLM-4-Plus foundation model has achieved parity with GPT-4o and other first-tier models in various metrics, said a spokesperson for ZhiPu. We are committed to making our models accessible to everyone, and we believe that our new models will have a significant impact on the development of artificial intelligence.
The release of these new models and the introduction of video calling on Qingyan App represent a significant step forward for ZhiPu AI. The company is positioning itself as a leader in the field of artificial intelligence, with a focus on developing and deploying cutting-edge technologies that can benefit both developers and end users.
Significanceof the Announcement:
This announcement holds significant implications for the Chinese AI landscape:
- Increased Accessibility: By offering free access to its GLM-4-Flash API, ZhiPu is making advanced AI technology more accessible to developers and researchers. This could accelerate the development of AI applications in China.
- C-End Focus: The launch of video calling on Qingyan App shows ZhiPu’s commitment to bringing AI benefits directly to consumers. This could lead to the development of new and innovative AI-powered consumer applications.
- Competition with International Players: ZhiPu’s claim of achieving parity with leading international modelslike GPT-4o indicates its ambition to compete with global giants in the AI space. This could drive further innovation and development in the Chinese AI industry.
Looking Ahead:
ZhiPu’s announcement highlights the rapid pace of development in the Chinese AI sector. The company’s focus on accessibility, consumer applications, and international competitiveness is likely to shape the future of AI in China. It remains to be seen how these new models will be adopted and integrated into various industries, but their potential impact is undeniable.
Views: 1