Question 1

Can RTX 3080 12GB run MPT-7B-Instruct?

Accepted Answer

Yes, RTX 3080 12GB can run MPT-7B-Instruct with a B grade (Runs well). Expected decode speed: 153.8 tok/s.

Question 2

How much VRAM does MPT-7B-Instruct need?

Accepted Answer

MPT-7B-Instruct (7B parameters) requires approximately 7.8 GB of memory with Q4_K_M quantization.

Question 3

What is the best quantization for MPT-7B-Instruct?

Accepted Answer

The recommended quantization for MPT-7B-Instruct is Q4_K_M, which balances quality and memory efficiency.

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	B	Runs well	153.8 tok/s	1831 ms	8K
Chat	C	Runs well	153.8 tok/s	687 ms	8K
Coding	B	Runs well	153.8 tok/s	1259 ms	8K
RAG	B	Runs well	153.8 tok/s	2289 ms	8K
Reasoning	B	Runs well	153.8 tok/s	1488 ms	8K

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	2.7 GB	Low	D33
Q3_K_S	3	3.4 GB	Low	D35
NVFP4	4

Can RTX 3080 12GB run MPT-7B-Instruct?

Memory breakdown