New comment by vafaeim in "GLM-5: Targeting complex systems engineering and long-horizon agentic tasks"

vafaeim — Thu, 12 Feb 2026 15:37:24 +0000

744B params is ~1.5TB VRAM (FP16). Even at int4, you need ~372GB just to load the weights (MoE sparsity saves FLOPs, not VRAM capacity). That's not a workstation, that's a rack with 5x H100s or a cluster of 8x RTX 6000 Adas.

The only real use cases here are strict data sovereignty (can't use US APIs) or using it as a teacher for distillation. Otherwise, the ROI on self-hosting is nonexistent.

Also, the disconnect between SOTA on Terminal bench and ~30% on Humanity's Last Exam suggests it overfitted on agent logs rather than learning deep reasoning.

Hacker News: vafaeim

New comment by vafaeim in "GLM-5: Targeting complex systems engineering and long-horizon agentic tasks"