Z.ai’s new GLM-4.5 AI model costs 11 cents per million input tokens and 28 cents per million output tokens, CNBC reported Monday (July 28), citing its interview with Z.ai CEO Zhang Peng.
For DeepSeek R1, those costs are 14 cents and $2.19, respectively, according to the report.
Z.ai’s GLM-4.5 is open source and can operate on eight Nvidia H20 chips, per the report.
The company said in a Monday blog post that the GLM-4.5 model series features two models — GLM-4.5 and GLM-4.5-Air — that satisfy the requirements of agentic applications by unifying reasoning, coding and agentic capabilities.
They are hybrid reasoning models that offer a “thinking mode” for complex reasoning and tool use, as well as a “non-thinking mode” for instant responses, according to the post.
“As a good [large language model (LLM)], it is necessary to deal with general problem solving, generalization, [commonsense] reasoning and self-improvement,” Z.ai said in the post. “In the past five years, OpenAI’s GPT-3 learns [commonsense] knowledge, and o1 uses reinforcement learning to think before [responding], significantly improving reasoning skills in coding, data analysis and complex math. However, the resultant models are still not really general. Some of them are good at coding, some good at math, and some good at reasoning, but none of them could achieve the best performance across all the different tasks. GLM-4.5 makes efforts toward the goal of unifying all the different capabilities.”
According to the CNBC report, Z.ai is on the United States’ entity list, meaning American companies are restricted from doing business with it.
The AI startup has raised $1.5 billion from investors and is planning an initial public offering (IPO) in China, per the report.
It was reported in March that Z.ai, then known as Zhipu AI, raised more than 1 billion yuan (about $140 million) in a funding round led by Hangzhou Municipal Construction Investment Group Co. and Shangcheng Capital, both of which have ties to the Hangzhou local government.
Existing backers Alibaba Group, Tencent and others also participated in the round.
Source: https://www.pymnts.com/news/artificial-intelligence