GLM-4.5-Air

Z.ai · 110.5B (active 12B) · MoE · ctx 128K · MIT · HuggingFace

GLM-4.5-Air (Z.ai) is a 110.5B parameters MoE (12B active per token) open-weight model tracked on AI Hashrate. At 8K context we estimate roughly 64.18 GB VRAM for Q4 (weights 60.78 GB + KV 2.4 GB + 1 GB overhead) and 224.4 GB for FP16. About 10 GPUs/accelerators in our catalog fully fit this model at Q4 under that context assumption. Fastest Q4 configs: B200 192GB SXM ≈ 424.2 tok/s (estimated); MI300X 192GB ≈ 281.1 tok/s (estimated); H200 141GB SXM5 ≈ 254.5 tok/s (estimated); H100 80GB SXM5 ≈ 177.7 tok/s (estimated); Gaudi 3 128GB ≈ 173.7 tok/s (estimated). Use the table below for VRAM fit and tok/s per dollar (list MSRP). Estimates follow memory-bandwidth math; measured rows override when present. See methodology for details. Methodology.

33 hardware configs — sorted by tok/s. Default fit context 8K.

Hardware	VRAM	Quant	tok/s	Fits?	tok/s/$
B200 192GB SXM	192	Q4	424.2 est.	Yes	0.011
MI300X 192GB	192	Q4	281.1 est.	Yes	0.019
H200 141GB SXM5	141	Q4	254.5 est.	Yes	0.007
B200 192GB SXM	192	FP16	244.0 est.	No	0.006
H100 80GB SXM5	80	Q4	177.7 est.	Yes	0.006
Gaudi 3 128GB	128	Q4	173.7 est.	Yes	0.014
MI300X 192GB	192	FP16	161.7 est.	No	0.011
A100 80GB SXM4	80	Q4	108.1 est.	Yes	0.007
RTX 6000 Ada 48GB	48	Q4	81.4 est.	No	0.012
H200 141GB SXM5	141	FP16	79.0 est.	No	0.002
L40S 48GB	48	Q4	73.2 est.	No	0.009
RTX 5090 32GB	32	Q4	67.5 est.	No	0.034
RTX A6000 48GB	48	Q4	65.1 est.	No	0.014
Gaudi 3 128GB	128	FP16	44.4 est.	No	0.004
Mac Studio M3 Ultra 96GB	96	Q4	43.4 est.	Yes	0.011
Mac Studio M2 Ultra 192GB	192	Q4	42.4 est.	Yes	0.006
Mac mini M4 Pro 64GB	64	Q4	41.1 est.	No	0.019
MacBook Pro M4 Max 128GB	128	Q4	28.0 measured	Yes	0.006
Mac Studio M2 Ultra 192GB	192	FP16	24.4 est.	No	0.003
RTX 4090 24GB	24	Q4	21.4 est.	No	0.013
RTX 4090D 24GB (China)	24	Q4	21.4 est.	No	0.013
RTX 3090 Ti 24GB	24	Q4	21.4 est.	No	0.011
MacBook Pro M3 Max 128GB	128	Q4	21.2 est.	Yes	0.005
RX 7900 XTX 24GB	24	Q4	20.3 est.	No	0.02
RTX 3090 24GB	24	Q4	19.8 est.	No	0.013
H100 80GB SXM5	80	FP16	17.7 est.	No	0.001
RX 7900 XT 20GB	20	Q4	11.8 est.	No	0.013
A100 80GB SXM4	80	FP16	10.8 est.	No	0.001
MacBook Pro M4 Max 128GB	128	FP16	7.4 est.	No	0.002
Mac Studio M3 Ultra 96GB	96	FP16	6.2 est.	No	0.002
MacBook Pro M3 Max 128GB	128	FP16	5.4 est.	No	0.001
Mac mini M4 32GB	32	Q4	4.5 est.	No	0.005
Mac mini M4 Pro 64GB	64	FP16	0.9 est.	No	0.0

Fit checks

• RTX 5090 32GB — does not fit · ≈67.5 tok/s Q4 est.
• RTX 3090 Ti 24GB — does not fit · ≈21.4 tok/s Q4 est.
• RTX 4090 24GB — does not fit · ≈21.4 tok/s Q4 est.
• RTX 6000 Ada 48GB — does not fit · ≈81.4 tok/s Q4 est.
• RX 7900 XTX 24GB — does not fit · ≈20.3 tok/s Q4 est.
• RTX 3090 24GB — does not fit · ≈19.8 tok/s Q4 est.
• L40S 48GB — does not fit · ≈73.2 tok/s Q4 est.
• Mac Studio M3 Ultra 96GB — fits · ≈43.4 tok/s Q4 est.
• Mac Studio M2 Ultra 192GB — fits · ≈42.4 tok/s Q4 est.
• RX 7900 XT 20GB — does not fit · ≈11.8 tok/s Q4 est.

12 popular retail GPUs; every row in the table above also links its Fits? verdict to the full fit check.

Same family: GLM

GLM-4.7 GLM-4.7-Flash
GLM-5 GLM-5.1 · GLM-5.2

All GLM models →