Microsoft has announced a new small language model named Fara-7B, designed to perform complex tasks directly on a user's device. Described as a Computer Use Agent (CUA) model, Fara-7B utilizes computer interfaces like the mouse and keyboard to achieve performance that the company claims is comparable to larger, more resource-intensive AI systems. This on-device processing approach offers significant advantages in terms of reduced latency and enhanced user privacy since data is stored and managed locally. The open-weight model has been made available on Microsoft Foundry and the Hugging Face platform under an MIT license. A version optimized for silicon is planned for future release via Copilot+ PCs running Windows 11.
Microsoft Announces Fara-7B, a Small Language Model for On-Device AI
MSFT
Related News
MSFT
OpenAI Nears Record $100B Funding at $850B Valuation to Fuel AI Infrastructure
MSFT
Microsoft Announces 240 New Marketplace Offers; Stock Up 1.06% Amid Broader Market Decline
MSFT
Reliance’s JioHotstar Taps OpenAI for India’s First ChatGPT-Powered Streaming Search
MSFT
Reliance to Invest $110 Billion in Sovereign AI and Green Data Centers
MSFT