(Yicai Global) July 24 -- Meta's recently updated large language model Llama 2 is an opportunity for developers as it is free to use but OpenAI's ChatGPT is still stronger, according to Chinese industry insiders.

Developers of LLM applications will be the first to benefit from the latest update of the Facebook operator's open-source product as they can provide LLMs to their customers at a low cost, boosting social productivity, Fu Sheng, founder and chairman of Cheetah Mobile, told Yicai Global. Cheetah is developing two LLMs, namely Chatmax.ai and Brainstorm.ai.

The emergence of complex free-to-use models is shaking up the artificial intelligence-generated content market kickstarted by Microsoft-backed OpenAI. Llama can be used for research and commercial purposes at no licensing cost.

"The quality of answers of Llama’s 70-billion-parameter version is already on par with ChatGPT 3.5," Fu said, adding that this means that many companies’ high investment in LLMs is eroding. OpenAI launched ChatGPT-4, the successor to ChatGPT 3.5, in March.

OpenAI still has its advantages as the San Francisco-headquartered company has more LLM servers than Llama, a developer working for a digital human firm said to Yicai Global.

"If a company uses OpenAI’s services, it can respond to 100 customers at the same time," the developer said, adding that a company would need more than 100 servers to achieve the same results with Llama 2 without even taking into account the cost of operation and maintenance.

Llama is mainly useful for teams that want to build their chatbots and AI agents based on the LLM, but it does not have many advantages for others, the developer added.

Chinese firms still need to make an effort in creating specific products, Chen Yunwen, chief executive of Data Grand, which recently unveiled an LLM called Cao Zhi, said to Yicai Global.

Editor: Emmi Laine