Nvidia’s new Llama-3.1 Nemotron Ultra outperforms DeepSeek R1

Posted under: AI technologies
Date: 2025-04-09
Llama 3.1: Nvidia's New Open-Source LLM | Justo Global

Nvidia has released a new open-source large language model (LLM) based on Meta's older model, Llama-3.1-405B-Instruct. The Llama-3.1-Nemotron-Ultra-253B-v1 is designed for advanced reasoning, instruction following, and AI assistant workflows. The model is designed to operate efficiently in both "reasoning on" and "reasoning off" modes, offering strong performance and cost-effective deployment in data center environments.

Read more at: venturebeat.com