TokenPost topic

Instinct MI355X

1 TokenPost story is tagged Instinct MI355X. The newest appear first.

Latest Instinct MI355X stories

• AMD’s MI355X Trails Nvidia B200 and B300 in AI Efficiency Test

  The MI355X delivered about 128 million tokens per second per megawatt at a target speed of 100 tokens per second per user, below both Nvidia systems.

  Oct 14, 2026, 1:23 PM EDT · Technology