Chinese startup Zhipu AI unveils a new large language model

Chinese artificial intelligence startup Zhipu AI has unveiled its latest foundational large language model-GLM4, as the Beijing-based company showcased the technological achievements it has achieved over the past three years.

Zhang Peng, CEO of Zhipu AI, said GLM-4 exhibits a significant overall performance boost, approaching the capabilities of GPT-4, a large language model developed by the US company OpenAI. Notably, it supports extended contextual understanding, boasts enhanced multi-modal capabilities, and achieves faster inference speeds, enabling higher concurrency and substantial cost reductions in reasoning.

Additionally, the personalized intelligent agent customization capabilities for GLMs are now live. Leveraging the robust capabilities of the GLM-4 model, users can create their own GLM intelligent agents with simple prompt commands, drastically reducing the entry barrier for large model utilization, the company said.

Meanwhile, Zhipu AI announced that it has initiated an open-source fund for large language models. This comprehensive initiative involves providing a thousand AI chip cards to the large language model open-source community, contributing 10 million yuan ($1.39 million) in cash to support open-source projects related to large language models.

Founded in 2019, Zhipu AI has received much attention in the past year. Last June, the US technology media The Information listed Zhipu AI as one of the five Chinese companies most likely to become China's OpenAI.