Zhipu AI's product, called AutoGLM Rumination, is capable of performing in-depth research as well as tasks like web searches, travel planning, and writing research reports, according to CEO Zhang Peng.
According to Zhipu AI's description, the GLM-Z1-Air model has comparable performance to rival DeepSeek's R1, but runs up to 8 times faster and requires only 1/30th of the computational resources.
The assistant is powered by Zhipu's proprietary models, including the GLM-Z1-Air inference model and the GLM-4-Air-0414 platform model. The company claims the GLM-Z1-Air has comparable performance to rival DeepSeek's R1 model, but runs up to eight times faster and requires only one-thirtieth of the computational resources.
AI assistants are systems designed to make decisions and perform a variety of tasks automatically. The launch follows a wave of AI product launches in China, starting with DeepSeek shaking up the industry earlier this year with a model that it said operated at a significantly lower cost than its US rivals.
The move comes just weeks after rival Manus grabbed headlines with what it touted as the world’s first general AI assistant. While Manus charges users up to $199 per month, Zhipu’s AutoGLM Rumination will be available for free through the company’s official channels, including the GLM model website and mobile app.
Zhipu AI, founded in 2019 as a spin-off from a Tsinghua University lab, has emerged as one of China's top AI startups. The company, which develops the GLM family of models, claims its latest large language model GLM-4 outperforms OpenAI's GPT-4 on several benchmarks.
The startup made headlines earlier this month after securing three rounds of government funding in a month, with the latest investment coming from the city of Chengdu, which poured 300 million yuan ($41.5 million) into the company.
On March 21, Chinese tech giant Tencent unveiled the final version of its T1 inference model, marking a new step forward in the increasingly fierce competition in the field of artificial intelligence (AI). According to Tencent's announcement on its official WeChat account, the upgraded T1 model offers faster response speeds and superior processing of long text documents.
The post said T1 can "keep clear content logic and neat, clean text," and that the rate of "hallucination" — a phenomenon where AI generates incorrect information — is "extremely low."
Earlier, Chinese internet search giant Baidu announced new artificial intelligence (AI) reasoning models on March 16 and decided to make its AI chatbot service free. In a WeChat post, Baidu announced that its latest X1 reasoning model - which the company claims has comparable performance to DeepSeek but at a lower cost - along with the Ernie 4.5 platform model, have been officially available on its AI chatbot Ernie Bot.
Source: https://www.baogiaothong.vn/cong-ty-trung-quoc-ra-mat-tro-ly-ai-mien-phi-chay-nhanh-gap-8-lan-so-voi-deepseek-192250331142015863.htm
Comment (0)