Question 1

Can AMD Instinct MI350X 288GB run Llama 3.3 70B?

Accepted Answer

Yes, AMD Instinct MI350X 288GB can run Llama 3.3 70B with a C grade (Runs well). Expected decode speed: 134.9 tok/s.

Question 2

How much VRAM does Llama 3.3 70B need?

Accepted Answer

Llama 3.3 70B (70B parameters) requires approximately 83.6 GB of memory with Q4_K_M quantization.

Question 3

What is the best quantization for Llama 3.3 70B?

Accepted Answer

The recommended quantization for Llama 3.3 70B is Q4_K_M, which balances quality and memory efficiency.

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	C	Runs well	134.9 tok/s	2464 ms	97K
Chat	C	Runs well	134.9 tok/s	924 ms	29K
Coding	C	Runs well	134.9 tok/s	1694 ms	55K
RAG	C	Runs well	134.9 tok/s	3080 ms	97K
Reasoning	C	Runs well	134.9 tok/s	2002 ms	55K

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	27.3 GB	Low	D31
Q3_K_S	3	34.3 GB	Low	D32
NVFP4	4

Can AMD Instinct MI350X 288GB run Llama 3.3 70B?

Memory breakdown