What is known thus far indicates that Huawei may release Pangu Chat, a rival to ChatGPT, in July.

 Huawei's large-scale learning model, which has 100 billion parameters, is the foundation of Pangu Chat.

It seems sense that more and more businesses would develop their own ChatGPT variants as the app becomes the preferred choice of millions of people. Known for its cutting-edge smartphones, Chinese tech giant Huawei is getting ready to introduce its own AI chatbot. At the moment, ChatGPT, Bard, Ernie bot, and many more AI chatbots are capturing the attention of users all over the world. It appears now that ChatGPT could soon have a strong challenger. 

Huawei is anticipated to introduce its ChatGPT competitor by July 7, according to the most recent rumours. Taking its name from the large-scale learning model used by the corporation, Huawei's latest AI software is referred to as Pangu Chat. At the Huawei Developer Conference (Cloud) conference, Pangu Chat will debut.Developed by Huawei, Pangu Chat

Top AI breakthroughs range from Chinese tech companies choosing generative AI to Germany potentially banning ChatGPT.

What are the details of Pangu Chat?

According to recent sources, Pangu Chat would initially only be accessible to businesses, and the company intends to sell it to the government and commercial users. Pangu Chat is said to be built on the expansive Pangu cloud model, which has 100 billion parameters. This number is less than OpenAI's GPT-3, which reportedly has over 175 billion parameters, and GPT-4, which is estimated to have over 1 trillion parameters. 

The skill of the model is defined by parameters, which are values. According to unsubstantiated sources, Pangu Chat's 100 billion parameters make it appear that it cannot compete with OpenAI's GPT-4. 

Too costly is ChatGPT Plus? 7 places where you can obtain GPT-4 without paying

According to reports, Huawei employed approximately 2000 Huawei Ascend 910 chipsets for more than two months of data training. According to market experts, the business annually trains a large-scale model on more than 4000 GPU/TPU cards. It's interesting to note that in just three years, Huawei's cost of computing power has surpassed 960 million yuan. 

Additionally, according to testing of the Pangu model, the Pangu-E big model produced close to 1.085 trillion parameters, which is comparable to the capacity of the GPT-4. The deep learning framework MindSpore is utilised in the creation of the Pangu-E large-scale model. The framework is comparable to GPT-3.5 in terms of dialogue generation.

Information on the Pangu Model

Huawei introduced the product in 2021, and the version 2.0 iteration was presented in April of the previous year. Currently, a huge model for natural language processing (NLP), a large model for weather, and a large model for CV power the model. 

NLP big models, scientific computing models, CV large models, etc. are all part of the Pangu Series AI big Models' top layer. The company's huge models that were created with partners make up the top layer.