Brave Browser Integrates RTX-Accelerated AI with Leo AI and Ollama

The Brave browser, known for its privacy focus, has introduced a powerful AI assistant, Leo AI, enhanced by RTX-accelerated local large language models (LLMs) through a collaboration with Ollama, according to the NVIDIA Blog. This integration aims to improve user experience by providing efficient, locally processed AI capabilities.

Enhanced AI Experience with RTX Acceleration

Brave's Leo AI, powered by NVIDIA's RTX technology, offers users the ability to summarize articles, extract insights, and answer questions directly within the browser. This is achieved through the use of NVIDIA's Tensor Cores, which are designed to handle AI applications by processing numerous calculations simultaneously. The collaboration with Ollama allows Brave to leverage the open-source llama.cpp library, which facilitates AI inference tasks specifically optimized for NVIDIA's RTX GPUs.

Advantages of Local AI Processing

Running AI models locally on a PC provides significant privacy benefits, as it eliminates the need to send data to external servers. This local processing approach ensures user data remains private and accessible without the necessity of cloud services. Additionally, it allows users to interact with various specialized models, such as bilingual or code generation models, without incurring cloud service fees.

Technical Integration and Performance

Brave's integration with Ollama and RTX technology offers a responsive AI experience, with the Llama 3 8B model achieving processing speeds of up to 149 tokens per second. This setup ensures quick responses to user queries and content requests, enhancing the overall browsing experience with Leo AI.

Getting Started with Leo AI and Ollama

Users interested in utilizing these advanced AI capabilities can easily install Ollama from its official website. Once installed, Brave's Leo AI can be configured to use local models through Ollama, offering flexibility to switch between cloud and local models as needed. Developers can explore more about using Ollama and llama.cpp through resources provided by NVIDIA.