According to Cointelegraph, Nvidia launched a new artificial intelligence model Llama-3.1-Nemotron-70B-Instruct on October 15, claiming that its performance surpassed GPT-4o and Claude-3.

According to a post on X.com from the Nvidia AI Developer account, the model performed well on lmarena.AI’s Chatbot Arena.

Llama-3.1-Nemotron-70B-Instruct is an improved version of Meta's open source model Llama-3.1-70B-Instruct, which Nvidia has optimized through specially curated datasets, advanced fine-tuning methods, and cutting-edge AI hardware.

Nvidia claims that the model scored 85 in Chatbot Arena’s “Hard” test, which, if true, would make it the top model in that test.

Llama-3.1-70B is Meta’s mid-range open source AI model, while GPT-4o is estimated to have used over a trillion parameters to develop.