Meta disrupted the race to enhance artificial intelligence capabilities last July by introducing Llama 2, an AI model similar to the one powering ChatGPT, available for public download and utilization. In November, a lesser-known startup from Beijing, 01.AI, unveiled its own open-source model surpassing Llama 2 in performance, ranking prominently on various leaderboards that assess AI model effectiveness.
Shortly after its debut, 01.AI’s model, Yi-34B, swiftly ascended to the top position on a ranking curated by startup Hugging Face, showcasing its prowess across standard benchmarks for automated intelligence. Subsequently, iterations of 01.AI’s model consistently achieved high scores on platforms like Hugging Face and other leaderboards. Recently, the startup introduced Yi-VL-34B, a “multimodal” AI model capable of analyzing images and engaging in discussions about their content.
While OpenAI, Google, and several AI enterprises maintain strict control over their technology, 01.AI is distributing its AI models freely, aiming to cultivate a devoted developer community to catalyze innovative AI applications. Founded in June last year, 01.AI has secured \(200 million in funding from Chinese ecommerce titan Alibaba and other investors, attaining a valuation exceeding \)1 billion.
The company’s CEO and founder, Kai-Fu Lee, a renowned investor with a background in groundbreaking AI research at Microsoft and Google, envisions Yi-34B as the culmination of his lifelong pursuit to create more intelligent machines. Emphasizing the importance of developing systems that comprehend human language, Lee envisions a future where computers understand speech and text, reflecting on the significance of bridging the gap between human communication and machine understanding.
Recognized as a frontrunner in China’s AI landscape, 01.AI seeks to lead the next phase of the AI revolution by pioneering “killer apps” leveraging the capabilities of language models. With plans to redefine productivity tools and explore AI-centric applications for various domains, the company aims to emulate the success of mobile-first platforms like Uber, WeChat, and TikTok, targeting global recognition and adoption.
Despite not launching any applications yet, 01.AI’s open-source language model has garnered acclaim in Western tech circles. Renowned AI expert Jerermy Howard lauds the model’s performance, surpassing even larger-parameter models in certain aspects, showcasing its potential for diverse AI applications.
Lee’s illustrious career in AI spans several milestones, from his pioneering research in speech recognition to his leadership roles at major tech companies like Apple, Microsoft, and Google. A staunch advocate for Chinese AI innovation, Lee’s efforts have contributed to the growth of China’s tech ecosystem, fostering collaborations and investments in AI startups and projects.
As the global AI landscape evolves, 01.AI’s open-source model, Yi-34B, represents a significant development, attracting interest and adoption beyond China. Positioned as a formidable contender in the AI domain, the company’s commitment to transparency and collaboration underscores its potential to drive innovation and advancements in AI technology.
