The current balance of power in open models
14 min
ZCode, the GLM coding agent, silently uploads your Git history
7 min2
GLM Built Its Own Inference Infrastructure
How Fragile Is Safety Alignment at Frontier Scale? A Single-Direction Attack on a 320B MoE
We have a year to fix security everywhere
13 min1
I tested 10 model/harness combinations on the same Three.js task
1
[AINews] 10% worse, 100x cheaper, 10000x faster: Why Simulation is taking over
15 min1