Question 1

How much VRAM does GLM-4.5-Air need?

Accepted Answer

GLM-4.5-Air (110.5B) requires approximately 246.9170648574829 GB of VRAM at full precision (FP16), 123.5 GB at 8-bit quantization (Q8), and 61.7 GB at 4-bit quantization (Q4). The actual VRAM usage may be slightly higher due to KV cache and runtime overhead.

Question 2

What GPUs can run GLM-4.5-Air?

Accepted Answer

We have 2 compatible GPUs indexed for GLM-4.5-Air, including MI325X, MI350X. Compatibility is determined by whether the GPU's VRAM can fit the model's memory footprint, potentially using quantization to reduce memory usage.

Question 3

What is the cheapest GPU to run GLM-4.5-Air?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with GLM-4.5-Air. Cloud pricing data is updated regularly — check back soon.

Question 4

Can GLM-4.5-Air run on consumer hardware?

Accepted Answer

GLM-4.5-Air requires approximately 246.9170648574829 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 61.7 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so GLM-4.5-Air requires datacenter or cloud GPUs.

Question 5

What quantization options are available for GLM-4.5-Air?

Accepted Answer

GLM-4.5-Air supports the following quantization levels: FP16 (full precision): ~246.9170648574829 GB, Q8 (8-bit): ~123.5 GB, Q6 (6-bit): ~92.6 GB, Q4 (4-bit): ~61.7 GB. Lower quantization reduces VRAM usage at the cost of some quality loss. Q8 offers a good balance between quality and memory savings, while Q4 provides maximum compression. As a Mixture of Experts (MoE) model with 128 experts (8 active), GLM-4.5-Air has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

GLM-4.5-Air

VRAM Requirements

Compatible GPUs

GLM-4.5-Air FAQ

VRAM requirements for GLM-4.5-Air at different quantization levels
Quantization	VRAM Required
FP16	246.9 GB
Q8	123.5 GB
Q6	92.6 GB
Q4	61.7 GB

Top 10 compatible GPUs for GLM-4.5-Air, sorted by cheapest price
GPU	VRAM	Best Quant	From	Rent
MI325X	256 GB	Q8	—
MI350X	288 GB	Q8	—