Kai-Fu Lee's AI startup 01.AI Launches Wanzhi AI Personal Assistant

Updated: May 13 2024 05:09

01.AI, led by the Kai-Fu Lee, has launched a one-stop AI work platform called 'Wanzhi'. This powerful tool is set to transform the way we work and interact with AI technology.

To ensure an unparalleled user experience, Kai-Fu Lee himself will serve as the Chief Experience Officer (CEO) of Wanzhi. In this role, he will actively listen to user feedback and work alongside everyone to unlock new and innovative uses for AI assistants. As part of this initiative, Lee will launch a series of tutorials showcasing how Wanzhi can be used to create new paradigms for work and life with the help of AI.



What sets Wanzhi apart from other general Q&A products is its ability to leverage various forms to output higher-quality content. By utilizing tables, simple mind maps, and other visually appealing formats instead of walls of text, Wanzhi makes it easier for users to understand and digest information. This feature is particularly useful in life-oriented scenarios, where Wanzhi can excel as a life assistant. With lightning-fast response times of less than a second, Wanzhi presents the required information in an easily digestible table format.


To address the potential lag in large model knowledge bases and prevent the generation of outdated or illusory information, Wanzhi supports real-time access and integration of Internet data. This feature ensures that users receive the most up-to-date insights and data, making Wanzhi a reliable and trustworthy AI assistant. Additionally, Wanzhi can process super long documents with hundreds of thousands of words within just 10 seconds, further demonstrating its impressive capabilities.



Alongside the launch of Wanzhi, 01.AI has also introduced its official API, allowing users to experience the Yi series AI large models. These models offer a range of capabilities suitable for various scenarios, such as chatting, Q&A, dialogue, collaboration, translation, massive data analysis, and visual reasoning. The optimized API interface significantly reduces model response latency, enhancing user experience and ensuring smooth interactions.

• Yi-Large: A high-performing, state-of-the-art LLM that excels in complex inference, prediction, and natural language understanding capabilities.





• Yi-Medium: A mid-size model with strong instruction following, code, and math capabilities. It offers balanced performance and high cost-effectiveness, making it suitable for general-purpose scenarios such as chat, Q&A, dialog, composing, and translation.





• Yi-Large-RAG: An augmented version of the Yi-Large model that integrates fresh know-how and knowledge. By leveraging retrieval and generation techniques, Yi-Large-RAG delivers precise answers and supports customers' private knowledge bases.

Yi-1.5 is an upgraded version of Yi. It is continuously pre-trained on Yi with a high-quality corpus of 500B tokens and fine-tuned on 3M diverse fine-tuning samples. In comparison with Yi, Yi-1.5 delivers stronger performance in coding, math, reasoning, and instruction-following capability, while still maintaining excellent capabilities in language understanding, commonsense reasoning, and reading comprehension. Yi-1.5 comes in 3 model sizes: 34B, 9B, and 6B. See Model Card for more details. Yi-34B-Chat and Yi-6B-Chat benchmark below:


With the launch of Wanzhi and the 01.AI API, Kai-Fu Lee and his team are ushering in a new era of AI-assisted work and life. Users can expect to see improvements in their productivity, efficiency, and overall quality of work. As Kai-Fu Lee personally guides users through tutorials on leveraging Wanzhi to create new paradigms for work and life, we can anticipate a future where AI and human collaboration becomes the norm, driving innovation and progress across various industries.

Yi-1.5-34B-Chat Space Web Demo: Chat UI Demo
Hugging Face: https://huggingface.co/01-ai
Github: https://github.com/01-ai
Paper: Yi: Open Foundation Models by 01.AI