Elon Musk's xAI has unveiled Grok-3, a significantly enhanced version of its Large Language Model (LLM). Announced during a livestream, Grok-3 runs on the Colossus supercluster which now boasts 200,000 GPUs, making it the largest of its kind. Grok-3 is designed to be an order of magnitude better than Grok-2, benefitting from extensive pre-training and innovative features such as the DeepSearch smart search engine. Available to Premium+ subscribers on X, initial benchmark tests suggest Grok-3 is outperforming its competition, achieving a groundbreaking score of 1400 on Chatbot Arena.
Grok 3 is an order of magnitude more capable than Grok 2.
xAI doubled the size of Colossus from 100,000 GPUs to 200,000 in just 92 days.
Grok-3 completed pre-training involving over 10x more training than its predecessor, Grok-2.
Grok-3 has reached a score of 1400 on Chatbot Arena, the first model to achieve such a rating.
Collection
[
|
...
]