Open-Source Lab Trains Near-3T Frontier Model

signulll · x · 2026-07-17

[Replying to @YuchenjUW] According to this reply, Kimi / Moonshot trained a frontier model approaching 3T parameters, achieving this under conditions involving fewer, restricted Nvidia GPUs.

The significance of this news goes beyond sheer scale; it emphasizes that frontier-level training is still achievable even under compute-constrained environments. The poster uses this case to illustrate that "innovation often stems from constraints."

Related event: Kimi K3 Triggers a Reassessment of Chinese Frontier AI(94 posts)→

Original post →

More from Models

Models channel →