China AI startup 01.AI valued at $1B in eight months

  • 01.AI aims to develop localised large language models for the mainland China market.
  • Achieved a $1 billion valuation after latest funding round from Alibaba Group.
  • LLM Yi outperforms established models like Llama 2, according to experts.

01.AI, the AI startup set up by Chinese computer scientist and CEO of Sinovation Ventures Kai-fu Lee, has reached a valuation of over $1 billion after a new funding round led by Alibaba Group’s cloud unit, just eight months after inception.

01.AI to develop localised large language models for mainland China market

OpenAI and Google are not available in mainland China. “Therefore, I believe many LLM professionals want to do their part to create a solution for those who truly need this market,” explained Kai-fu Lee.

The growth of AI reflects the rapid development of the field of generative artificial intelligence. Seven months after its establishment, this startup company released its first bilingual (Chinese-English) open-source model, called Yi.

Also Read: China’s Baichuan Intelligence Unveils Open Source Language Model to Compete with OpenAI

Open-source Yi in two versions, outperforms established AI models

Yi includes two versions: Yi-34B and Yi-6B. Yi-6B is best suited for personal use and research, while Yi-34B has greater power for meeting the needs of the open source community. Both are currently open for free commercial applications.

According to Hugging Face, a leaderboard that ranks and compares different AI models, Yi-34B is a bilingual basic model trained with 34 billion parameters, much smaller than other open models such as Falcon-180B and Meta LlaMa2-70B. However, Yi-34B has a data parameter volume of 34 billion, can support 200K ultra long context windows, and can handle approximately 400,000 Chinese characters and ultra long text input, making it the longest version in the world at present. “The decision to launch an open LLM as the first product is a ‘feedback’ to society. We have provided a trustworthy choice,” added Kai-fu Lee.

Also Read: Meta Launches Game-Changing Open-Source AI Model, LLaMA 2, Free for All Users

Establishing an LLM is a costly undertaking. In order to maintain cash intensive operations, 0.AI was developed using a commercial fee plan. Although the company will continue to open source some of its models, its goal is to establish a state-of-the-art proprietary model as the foundation for various commercial products.

“AI 2.0 is the largest technological revolution in history,” said Kai-Fu Lee, pointing to the scale of the project he has taken on.

Vicky-Huang

Vicky Huang

Vicky Huang was a reporter at btw media. She majored in international journalism and communication at Beijing Foreign Studies University.

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *