• Riding the generative AI trend, vivo has established the ‘1+2+N’ matrix to explore new possibilities on smartphones.
• In 2024, various generative AI use cases will be applied in both flagship and more affordable high-end smartphones by combining scalable AI capabilities from cloud to edge.
• 40% of smartphones shipped in 2027 will be capable of locally running a number of generative AI applications.

vivo launched the BlueLM along with other new technology pillars during the 2023 vivo Developer Conference held on November 1. The self-developed AI big model can provide comprehensive and scalable AI capabilities from cloud to edge, with the number of parameters ranging from 175 billion to 1 billion.

vivo also announced the establishment of the ‘1+2+N’ matrix, which has BlueLM as its backbone. Here, the ‘2’ refers to the enhanced voice assistant ‘Blue Heart V’ and the robot app “Blue Heart Qianxun” to execute natural language conservation, while the ‘N’ represents unlimited possibilities enabled by specially tailored models. These custom models normally demand less computing and memory and are more efficient and more affordable when running a specific task.

On December 21, vivo held a press conference at its new headquarters in Dongguan, China, where the BlueLM was again the highlight. During the event, vivo showcased several generative AI applications powered by BlueLM. With the vivo X100 or X100 pro, AI can be instructed to write a review, generate a transcript for a blog post or summarize an article, and even draw a picture with a simple description. It also enables a smarter voice assistant with a significant improvement in understanding human speech, thus providing access to desired services and content more efficiently.

Moreover, BlueLM on the cloud has hundreds of billions of parameters. When chatting with the ‘Blue Heart Qianxun’, conservations become more natural and meaningful, and we can soon expect more accurate and informative answers to myriad questions.

The new wave of technological advancement is also in favor of visually impaired individuals. There were demos where AI recognized different objects with the rear camera and vocally described them. The function can be also used to find people and animals. The move resonates with and echoes vivo’s commitment to technology sharing.

The above-mentioned demonstration is just the tip of the iceberg of the promise of generative AI, which is bound to elevate the smartphone experience. Commenting on its development, Counterpoint Senior Analyst Bai Shenghao said, “In 2024, top OEMs will continue to explore new possibilities atop in-house big models, and the local implementation of various generative AI applications will become a key differentiator for premium smartphones. Meanwhile, part of practical use cases will be introduced to more affordable models by connecting to the cloud, so more smartphone users can help train the big models.” According to Counterpoint’s latest forecasts on generative AI smartphone penetration, 40% of smartphones shipped in 2027 will be capable of locally running several generative AI applications.

Currently, Chinese OEMs attach greater importance to the Large Language Model (LLM). Over time, smartphone interaction will become more natural and intuitive, and the sufficiently-trained LLM will be able to better interpret texts, voice and images, and greatly improve smartphone productivity. Bai added, “The combination of foldable and generative AI technologies will make the smartphone an indispensable mobile productivity powerhouse in the near future.”