Chinese artificial intelligence (AI) company Moonshot AI has halved the price of a new feature of its large language model (LLM) Kimi, as startups compete with the country’s tech giants to monetize their generative AI products.
The price of Moonshot’s Kimi context caching feature has been reduced from 10 yuan to 5 yuan ($0.70) per 1 million tokens per minute, Moonshot said in a blog post on Wednesday.
Context caching allows LLM developers to store certain potentially frequently requested information for a period of time so that their model responds more quickly to similar queries.
The feature could “significantly increase efficiency and reduce costs,” Moonshot said when it launched a public test of the feature in July.
Moonshot is the latest generative AI company in China to lower its service prices amid fierce competition in the country, where major tech companies and startups are racing to commercialize their LLMs, the technology that underlies generative AI services like ChatGPT.
Online search engine giant Baidu, TikTok owner ByteDance and social media and video game giant Tencent Holdings have all cut prices on their LLM products, creating headwinds for the numerous AI startups in China that are also trying to make profits.
Moonshot, which is backed by investors such as Alibaba, is part of a group of well-known Chinese AI start-ups that have emerged as potential challengers to the tech giants and OpenAI in the US.
Founded in 2019 to build on research from the Knowledge Engineering Group at Beijing’s prestigious Tsinghua University, Beijing Zhipu Huazhang Technology is another closely watched domestic AI unicorn, defined as privately held startups valued at more than $1 billion.
In June, Zhipu AI also announced that it would reduce the price of its GLM-LLM series by 90 percent. According to the company, this is not just a simple price war, but shows its innovative strength.
