Nvidia unveils new GPU designed for long-context inference

At the AI ​​Infrastructure Summit on Tuesday, Nvidia announced a new GPU Rubin CPXThe reference is designed for windows that are larger than 1 million tokens.

Chip is adapted to process larger sequences of CPX references, part of the upcoming Rubin series of the giant and is used to use it A comprehensive “disagreeable conclusion” infrastructure approachFor users, the result will be better on long -term functions such as video generation or software development.

The tireless development cycle of NVIDIA has resulted in huge advantage for the company, which has been brought in $ 41.1 billion in the sale of data centers. Its most recent quarter,

Rubin CPX is slapped to be available at the end of 2026.

Source link

Please follow and like us:
Pin Share

Leave a Reply

Your email address will not be published. Required fields are marked *