How much VRAM does GLM-4 9B need?

GLM-4 9B (9B parameters) requires approximately 9.3 GB of memory with Q4_K_M quantization.

What is the best quantization for GLM-4 9B?

The recommended quantization for GLM-4 9B is Q4_K_M, which balances quality and memory efficiency.

Can it run?

Yes, Intel Arc B580 12GB can run GLM-4 9B with a C grade (Runs well). Expected decode speed: 59.8 tok/s.

CUsable

Runs well

Using Q4_K_M in Ollama

Capabilities:

Fit status

Runs well

Decode

59.8 tok/s

TTFT

3237 ms

Safe context

21K

Memory

9.3 GB / 12.0 GB

Weights5.5 GB

KV Cache1.4 GB

Runtime1.2 GB

Headroom1.2 GB

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	C	Tight fit	39.9 tok/s	7063 ms	36K
Chat	C	Runs well	39.9 tok/s	2649 ms	11K
Coding	C	Runs well	59.8 tok/s	3237 ms	21K
RAG	C	Tight fit	39.9 tok/s	8829 ms	36K
Reasoning	C	Runs well	39.9 tok/s	5739 ms	21K

How GLM-4 9B (9B params) fits at each quantization level on Intel Arc B580 12GB (12.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	3.5 GB	Low	D35
Q3_K_S	3	4.4 GB	Low	D37
NVFP4	4

Ollama

ollama run glm-4-9b

HuggingFace

huggingface-cli download glm-4-9b

Upgrade options

RX 9070 XT 16GBBudget pick

C74.6 tok/s decode

~$549 MSRP

RTX 5070 Ti 16GBBest value

C104.5 tok/s decode

~$749 MSRP

RTX 4070 Ti Super 16GBBiggest leap

C97.9 tok/s decode

~$799 MSRP