Can these new Chinese AI models dethrone OpenAI? Find out

When it comes to technology, China always seems to be in a competitive mode and comes up with developments that put it one step ahead on the global map. And this time the Chinese have their eyes set on artificial intelligence models. Currently, Open AI rules globally with its ChatGPT and has the first-mover advantage. But we now hear that China has developed artificial intelligence models that could beat OpenAI on various fronts. Let’s find out more about it and answer: can these models be the future of the AI industry?

The Beijing Lingyi Wanwu Information Technology Company has built the Yi series, which is a progressive generative chatbot. It was developed in its 01.AI lab. Further, this is a large language model (LLM) and comes in two versions- the lightweight Yi-6B-200K and the more robust Yi-34B-200K. Both of these can understand both English and Mandarin and can retain immense conversational context.

Also read: Sundar Pichai: Five quotes on 25 years of Google, future of search and AI

Hugging Face, which is a key repository for AI models, has ranked the Yi model as the second most popular open-source model, within hours of its release. Even on several synthetic benchmarks, Yi models beat other LLMs and proved to be very efficient and accurate. Further, o1.AI also shared the scoreboard of the Yi models which reflected that they have a really strong performance in aspects such as reading comprehension and common-sense reasoning. The two also performed strongly on common AI tests like Gaokao and C-eval.

01.A1 said, “Yi-34B outperforms much larger models like LLaMA2-70B and Falcon-180B; also Yi-34B’s size can support applications cost-effectively, thereby enabling developers to build fantastic projects.”

Also read: Satya Nadella retweets Microsoft President’s blueprint for India to govern AI

Can the Yi models beat OpenAI models?

There are two major reasons that weigh in towards this possibility:

1. The fact that this is a fully open-sourced model means that anyone can run them locally on their system, allowing for greater control, the ability to modify the model architecture, and avoiding reliance on external servers.
2. So, large language models like ChatGPT and Yi Series process units of text or “tokens” while working. This could even mean the smallest part of a word.

So coming back to point 2, the Yi series can process up to 200,000 tokens of context in a single go. This far beats OpenAI’s GPT-4 Turbo which processes 128,000 tokens and even Anthropic’s Claude which deals with 100,000 tokens.

Let’s now see how this Yi series progresses in the AI world.