ChipsAug 24, 2026 · 23:528 sources

Xiaomi Xring O100 Hits 1.22 TB/s Bandwidth, First Dedicated Edge AI Accelerator

From huggingnews.com

Image: huggingnews.com
AI;DR

Xiaomi's Xring O100 is a 6nm edge AI accelerator that uses 3D wafer-on-wafer stacking to deliver 1.22 TB/s of near-memory bandwidth. The chip processed a 3B parameter model at 330 tokens per second in lab tests and is scheduled for commercial use in 2027. Xiaomi is deploying the hardware in an AI Cube prototype to run models up to 120B parameters at 150W sustained power.

Read the sourceSign in to suggest

Sources

+ 6 more
  1. SUPPORT@jiemian_news“can run Xiaomi’s MiMo 3B model at 330 tokens/s”x.com
  2. SUPPORT@zephyr_z9“This spec is extremely similar to Huawei's Logic Folding”x.com
  3. SUPPORT@maxforai“D100 is Xiaomi's smart driving high-performance AI chip, supporting up to 160GB of memory and 200B large model local deployment”x.com
  4. SUPPORT@teortaxestex“6nm, Wafer-on-wafer NPU-DRAM bonding with 1.4µm pitch”x.com
  5. SUPPORT@tonyjzhou“Decoding re-reads active weights every token, so local inference is bandwidth-bound, not FLOPs-bound”x.com
  6. SUPPORT@hesamation“Xiaomi's new prototype looks like a direct competitor to DGX Spark”x.com

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today