Question 1

Can AMD Instinct MI350X 288GB run Llama 4 Maverick 17B 128E?

Accepted Answer

Yes, AMD Instinct MI350X 288GB can run Llama 4 Maverick 17B 128E with a C grade (Runs with offload). Expected decode speed: 70.7 tok/s.

Question 2

How much VRAM does Llama 4 Maverick 17B 128E need?

Accepted Answer

Llama 4 Maverick 17B 128E (400B parameters) requires approximately 277.9 GB of memory with Q4_K_M quantization.

Question 3

What is the best quantization for Llama 4 Maverick 17B 128E?

Accepted Answer

The recommended quantization for Llama 4 Maverick 17B 128E is Q4_K_M, which balances quality and memory efficiency.

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	C	Runs with offload	71.7 tok/s	3930 ms	33K
Chat	C	Runs with offload	71.7 tok/s	1474 ms	8K
Coding	C	Runs with offload	70.7 tok/s	2739 ms	17K
RAG	C	Runs with offload	71.7 tok/s	4912 ms	33K
Reasoning	C	Runs with offload	71.7 tok/s	3193 ms	17K

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	156.0 GB	Low	C41
Q3_K_S	3	196.0 GB	Low	C44
NVFP4Best for your GPU

Can AMD Instinct MI350X 288GB run Llama 4 Maverick 17B 128E?

Memory breakdown