Question 1

Can NVIDIA H100 80GB run DeepSeek LLM 67B?

Accepted Answer

Yes, NVIDIA H100 80GB can run DeepSeek LLM 67B with a B grade (Runs well). Expected decode speed: 59.0 tok/s.

Question 2

How much VRAM does DeepSeek LLM 67B need?

Accepted Answer

DeepSeek LLM 67B (67B parameters) requires approximately 61.7 GB of memory with Q4_K_M quantization.

Question 3

What is the best quantization for DeepSeek LLM 67B?

Accepted Answer

The recommended quantization for DeepSeek LLM 67B is Q4_K_M, which balances quality and memory efficiency.

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	C	Tight fit	68.9 tok/s	4090 ms	4K
Chat	B	Runs well	68.9 tok/s	1534 ms	4K
Coding	B	Runs well	59.0 tok/s	3280 ms	4K
RAG	C	Tight fit	68.9 tok/s	5112 ms	4K
Reasoning	B	Runs well	68.9 tok/s	3323 ms	4K

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	26.1 GB	Low	D36
Q3_K_S	3	32.8 GB	Low	D38
NVFP4	4

Can NVIDIA H100 80GB run DeepSeek LLM 67B?

Memory breakdown