Qualcomm says on-device AI models will offer enhanced privacy
On-device AI capability will also reduce latency in processing AI tasks
Currently, Qualcomm AI Hub offers Mistral 7B v0.3 AI model
The AI models will be optimised for Snapdragon 8 Elite, Cockpit Elite, Ride Elite, and X Elite platforms
Photo Credit: Reuters
Qualcomm announced a partnership with artificial intelligence (AI) firm Mistral on Wednesday. Announced at the Snapdragon Summit, this partnership will see Mistral's two AI models — Ministral 3B and Ministral 8B — be optimised to run on-device on devices powered by various Snapdragon Elite series chipsets. The company said that these AI models will add AI capabilities to smartphones, PCs, vehicles, and more. The move is aimed at improving processing speeds and reducing the power consumption of AI tasks.
Qualcomm, Mistral Join Hands to Introduce On-Device AI Models
In a press release, the chip maker formally announced that Qualcomm and Mistral will be working together to introduce specific AI models designed to run on-device. The move is aimed at reducing the reliance on edge and cloud processing of AI tasks, which is considered both energy-intensive and slower compared to local processing. Qualcomm also added that these AI models will be optimised for not only its Mobile Platforms but also for PCs and the vehicle ecosystem.
With this partnership, Ministral 3B and Ministral 8B will be optimised to run on devices powered by the recently launched Snapdragon 8 Elite for smartphones, Snapdragon Cockpit Elite and Snapdragon Ride Elite for automotives, and Snapdragon X Elite for AI PCs.
The Ministral 3B and 8B AI models were launched last week. They are the company's first large language models (LLMs) designed for commercial usage. These models support function-calling making them ideal for application programming interface (API) use cases. As per Mistral, the AI models support up to 1,28,000 token context length.
Qualcomm stated that on-device AI models will offer multiple benefits to end users including improved privacy, lower latency, higher reliability, energy efficiency, and cost savings. “Mistral AI's Ministral 3B and Ministral 8B will enable device manufacturers, software vendors, and digital service providers to deliver innovative experiences, such as AI assistants and other applications that understand users' wants and need,” said Durga Malladi, Senior Vice President and General Manager of Technology, Planning and Edge Solutions at Qualcomm Technologies.
Currently, developers can access the Mistral 7B v0.3 AI model via the Qualcomm AI Hub. The two Ministral AI models will be added to the hub in coming weeks.
OneXSugar With Snapdragon G3 Gen 3, Dual Screens Unveiled Alongside Ayaneo Pocket S2, Ayaneo Gaming Pad at GDC 202518 March 2025
Search
