Nvidia unveils new GPU designed for long-context inference

Nvidia declared a new GPU at the AI ​​infrastructure conference on Tuesday Rubin CPXDesigned for Windows in the context larger than 1 million token.

Part of the upcoming Rubin series of the chip giant, the larger sequences of the CPX context have been imitated for processing and it is meant to be used as part of it A broad “individual estimate” infrastructure approachThe For users, the result will be better performance in long-tasks such as video generation or software development.

Nvidia’s relentless development cycle has caused a lot of profit for the organization, which brings the data center sales $ 41.1 billion This is the most recent marginalThe

Rubin will be available at the end of CPX 2026.