Question 1

Can RTX A5000 24GB run GPT-OSS 20B?

Accepted Answer

Yes, RTX A5000 24GB can run GPT-OSS 20B with a B grade (Runs well). Expected decode speed: 98.7 tok/s.

Question 2

How much VRAM does GPT-OSS 20B need?

Accepted Answer

GPT-OSS 20B (21B parameters) requires approximately 17.2 GB of memory with Q4_K_M quantization.

Question 3

What is the best quantization for GPT-OSS 20B?

Accepted Answer

The recommended quantization for GPT-OSS 20B is Q4_K_M, which balances quality and memory efficiency.

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	B	Runs well	96.0 tok/s	2933 ms	44K
Chat	B	Runs well	96.0 tok/s	1100 ms	11K
Coding	B	Runs well	98.7 tok/s	1961 ms	22K
RAG	B	Runs well	96.0 tok/s	3667 ms	44K
Reasoning	B	Runs well	96.0 tok/s	2383 ms	22K

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	8.2 GB	Low	D37
Q3_K_S	3	10.3 GB	Low	D38
NVFP4	4

Can RTX A5000 24GB run GPT-OSS 20B?

Memory breakdown