The startup's offerings are attractive for global organizations serving customers in China, as they can use the model to build bilingual chatbots. The company plans to add support for more languages to the open-source models and launch a larger commercial LLM targeting OpenAI’s GPT series. Other companies, such as Baidu, Naver, and India’s Reliance Industries, are also focusing on developing LLMs for specific languages and markets.
Key takeaways:
- 01.AI, a Chinese startup founded by Kai-Fu Lee, has released a 34-billion parameter large language model (LLM) called Yi-34B that outperforms the 70-billion Llama 2 and 180 billion Falcon models built by Meta Platforms, Inc., and the Technology Innovation Institute in Abu Dhabi, respectively.
- The new AI model supports Chinese and English languages and can be fine-tuned for a variety of use cases. The startup also offers a smaller model trained with 6 billion parameters.
- 01.AI plans to double down these models and launch a commercial offering capable of taking on OpenAI, the current generative AI market leader by number of users.
- Other companies such as Baidu, Naver, and India’s Reliance Industries are also focusing on developing large language models for specific languages and markets.