DEC 14 (Reuters) – Beijing/Hong Kong Building on a $1 billion valuation attained last month due to a surge in interest in open-source AI models worldwide, Beijing-based firm 01.AI is reportedly seeking up to $200 million in new funding, according to two people familiar with the situation.
Along with bigger companies Meta (META.O) and Alibaba (9988.HK), it is one of several Chinese businesses that have opened, or plan to publish, their large language models (LLM) for public usage in an attempt to catch up to industry leader OpenAI.
According to one of the sources, 01.AI, which was formally introduced by Google China’s previous CEO, Lee Kai-fu, in July following a three-month incubation phase, reached a $1 billion valuation early in November.
According to the other source, the business was looking for more investors denominated in US dollars.
Since the specifics of the funding were kept confidential, every source referenced in this piece requested anonymity.
A request for comment was not immediately answered by the corporation.
When 01.AI’s Yi-34B model topped Huggingface’s leaderboard in November—a website that allows tech companies to exchange LLMs and score them based on popularity and performance—it became the first Chinese LLM to receive recognition in the open-source LLM community.
For instance, Zhipu AI, which was formed in 2019, is valued more than $1 billion, according to Chinese media.
After its most recent fundraising in October, eight-month-old Baichuan Technology reached a valuation of $1.2 billion, according to a third source with intimate knowledge of the operation.
THE OPEN-SOURCE DEBATE
OpenAi, the company behind the popular chatbot ChatGPT, has criticized the shift to open-source LLMs, citing concerns about potential manipulation by hostile actors who could endanger society. The company has kept the codes of its models tightly confidential.
Tech behemoth Alibaba’s cloud division (9988.HK) in China has been actively releasing its LLMs as open-source. The second model from China to top Huggingface’s leaderboard was its most recent LLM Qwen-72B.
Alibaba has released eight AI models in the last four months, the most recent and potent of which is Qwen-72B. According to the company, Qwen-72B can handle Chinese better than OpenAI’s flagship GPT4 based on some industry benchmarks.
When questioned about Alibaba’s commitment to open-sourcing its AI models, Jingren Zhou, the chief technical officer of Alibaba Cloud, told Reuters on Wednesday that “building up an open-source ecosystem is critical to promoting the development of LLM and AI applications building.”
According to him, there is an increasing demand for AI applications from developers, businesses, and a variety of industries.
“Alibaba Cloud aspires to become the most open cloud and make generative AI capabilities accessible to everyone.”
Additionally, the business supports smaller open-source AI LLM businesses, such 01.AI, with whom it works on model training and deployment procedures, among other things.
The sources indicated that earlier in the current round, AliCloud pledged funding.
A request for comment on Thursday regarding the 01.AI funding was not immediately answered by AliCloud.
