Foreign media: Chinese AI company MiniMax has released a new M2.5 model, claiming it is the "first frontier model that users don't need to worry about costs," with an hourly usage cost of only $1, capable of processing 100 tokens per second.
The model has a parameter scale of 230 billion, the same as the previous M2 model, but its performance is comparable to leading models such as Anthropic and OpenAI in the United States, especially excelling in programming and search.
At the same time, MiniMax's competitor Zhipu AI also launched a new GLM-5 model with a parameter scale of 744 billion. This shows the rapid development of China's AI industry in technological innovation and cost control, attracting the attention of global investors.
Original article: toutiao.com/article/1857006101447687/
Statement: This article represents the views of the author himself.