Chinese startup Zhipu is developing an AI model capable of competing with DeepSeek.

Chinese AI manufacturer Zhipu is preparing to launch a new model GLM‑5

- What’s expected?

In the near future the company plans to unveil an updated version of its flagship language model – GLM‑5. Forecasts suggest it could compete with DeepSeek’s major release, which is also slated for the coming months.

- Technical details

GLM‑5 is designed to tackle complex programming tasks and AI agent management. During testing it was compared with Anthropic Opus models, and in terms of parameters it doubles its predecessor. The official launch is scheduled for February 12.

- Industry context

Ahead of Chinese New Year new open models have already appeared: Moonshot and ByteDance.

- *Moonshot* released its model in January, which now holds leading positions among open‑source projects.

- *ByteDance* introduced the video model Seedance 2, receiving high praise for its broad functionality.

DeepSeek, according to unverified reports, plans to announce a major update right during the celebration, which could become a new benchmark for Chinese open AI models.

- Brief history of Zhipu

- The first large Chinese AI company to go public – the company’s valuation is $18 billion.

- Founded in 2019 under the leadership of scholars from Tsinghua University; the project received support from state funds and private capital (Alibaba, Tencent).

- Today Zhipu is shifting from developing customized solutions for businesses to services for a wider audience: subscriptions for AI programming are already available.

- Internal comment

Tan Jie, co‑founder and chief scientific officer of the company, announced the imminent launch of GLM‑5 and emphasized the need to return to fundamental research. He noted: “DeepSeek has become an alarm bell for us.”

Thus, Zhipu is preparing a powerful model that could reshape the landscape of open AI solutions in China while responding to competitor activity from DeepSeek, Moonshot, and ByteDance.