GLM-4.5-Air: the GLM you can actually run at home (hardware guide)

GLM-4.5-Air is 106B total but only 12B active, ~60GB at 4-bit — small enough for one 80GB card or a 128GB unified box. Exactly what hardware runs… Aliteq Lena Fischer · AI & Local Compute Editor

106B total / 12B active, 128K context (HF model card, 24 Sep 2026).

~60GB at 4-bit — fits one 80GB card (A100/H100), a 96GB RTX PRO 6000, or a 128GB unified box.

Unified memory works well: a Strix Halo or Mac with 128GB holds it with context to spare.

MoE = fast for its size — only 12B active per token despite 106B loaded.

No 60GB+ machine? Rent an 80GB card from about $0.47/hr.

Read the full story

Aliteq GLM-4.5-Air: the GLM you can actually run at home (hardware guide) Read the full story on Aliteq