MediaTek touted last year the AI capabilities of its upcoming Dimensity 9300 and 8300 chipsets. Today, that development takes another significant leap. Because the company has just announced its collaboration with Google, that would allow these SoCs to work optimally with Gemini Nano. This joint project aims to make Generative AI features more accessible to a wider range of smartphones starting in 2024.
Dimensity 9300 and 8300 AI Perks Recap
The Dimensity 9300 and 8300 SoCs were designed with a focus on enabling developers and OEMs to provide key AI capabilities directly to users on-device. Generative AI, edge AI processing, and AI camera enhancements, among other things. This in turn provided features such as image recognition, voice recognition, and natural language processing. MediaTek's AI technology also enhanced gaming experiences by providing features like AI-based game enhancements and real-time performance optimization.
Gemini Nano Optimization
According to the official press release, the optimization using Gemini Nano involved integrating and optimizing Google's Large Language Model (LLM). This integration allows Gemini Nano to run effectively and efficiently on the aforementioned chipsets, enhancing their AI capabilities.
This project even included MediaTek's NeuroPilot toolkit and porting Gemini Nano to MediaTek's APU to improve performance. By optimizing Gemini Nano for these chipsets, MediaTek aims to enable developers and OEMs to include Generative AI features into their applications by default.
Why is this Important?
One of the key advantages of on-device Generative AI is its ability to provide seamless performance, greater privacy, better security and reliability, lower latency, offline functionality, and lower operation costs when compared to cloud computing. This optimization allows developers and OEMs to leverage these benefits and deliver enhanced AI experiences to end-users.
The Dimensity 9300 chip, in particular, is expected to play a crucial role in MediaTek's market share growth, powering devices like the Vivo X100 series and the Oppo Find X7. Its performance has been praised, with some analysts considering it the most powerful smartphone SoC currently in use.
MediaTek and Google’s collaboration also involves the development of an APK that will aid developers and OEMs in deploying Gemini Nano applications on the Dimensity 9300 and 8300 SoCs. This APK will enable easier integration of Gemini Nano into apps, further expanding the reach of on-device Generative AI.
Advanced AI for the Masses?
Of course, we cannot dismiss this development's importance to the Dimensity 8300, which can definitively bring generative AI capabilities to the mid-range 5G smartphone market. Flagship devices like the Google Pixel 8 Pro and Samsung Galaxy S24 series currently benefit from Gemini Nano, powering features such as Smart Reply and summaries in the Recorder app. However, with MediaTek's optimization, more devices, including mid-range phones like the Poco X6 Pro and OPPO Reno 12, will now have access to these AI capabilities.
With such a development, these phones would not only offer a balance of performance and affordability, but effectively become an easy gateway for any user to access advanced AI features. Which then hopefully makes AI technology even more prevalent and universal to society.
