How much VRAM does Qwen 3 235B A22B need?

Qwen 3 235B A22B (235B parameters) requires approximately 157.3 GB of memory with Q4_K_M quantization.

What is the best quantization for Qwen 3 235B A22B?

The recommended quantization for Qwen 3 235B A22B is Q4_K_M, which balances quality and memory efficiency.

Can it run?

No, Qwen 3 235B A22B requires more memory than NVIDIA GH200 96GB provides.

FWon't run

Too heavy

Using Q4_K_M in llama.cpp

Capabilities:

Fit status

Too heavy

Decode

53.6 tok/s

TTFT

3613 ms

Safe context

10K

Memory

157.3 GB / 96.0 GB

Weights143.4 GB

KV Cache3.4 GB

Runtime0.9 GB

Headroom9.6 GB

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	F	Too heavy	53.6 tok/s	5255 ms	19K
Chat	F	Too heavy	53.6 tok/s	1971 ms	5K
Coding	F	Too heavy	53.6 tok/s	3613 ms	10K
RAG	F	Too heavy	53.6 tok/s	6568 ms	19K
Reasoning	F	Too heavy	53.6 tok/s	4269 ms	10K

How Qwen 3 235B A22B (235B params) fits at each quantization level on NVIDIA GH200 96GB (96.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	91.7 GB	Low	C45
Q3_K_S	3	115.2 GB	Low	F0
NVFP4	4

Upgrade options

AMD Instinct MI350X 288GBBudget pick

B108.7 tok/s decode

~$8,000 MSRP

AMD Instinct MI300X 192GBBest value

C76.9 tok/s decode

~$15,000 MSRP