Intel · 128 GB HBM2e · 3276 GB/s bandwidth ·
Z.ai · 110.5B (active 12B) · MoE · model ctx up to 128K
Yes — the Gaudi 3 128GB (128 GB HBM2e) can run GLM-4.5-Air. At Q4 with the default 8K context it needs ≈ 64.18 GB VRAM (weights 60.78 GB + KV cache 2.4 GB + 1 GB runtime overhead), which fits within 128 GB. Decode at Q4/8K: ≈ 173.7 tok/s (estimated, batch 1). FP16 needs ≈ 224.4 GB — does not fit on this card. Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.
Fit & speed by quant and context
Quant
Context
VRAM needed
Fits?
tok/s (decode)
Q4
4K
62.98 GB
Yes
173.7est.
Q4
8K default
64.18 GB
Yes
173.7est.
Q4
32K
71.38 GB
Yes
173.7est.
FP16
4K
223.2 GB
No
44.9est.
FP16
8K default
224.4 GB
No
44.4est.
FP16
32K
231.6 GB
No
41.7est.
Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured
anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s
means the model is far beyond this card (offload-only territory).