Question 1

Can RTX 4090 24GB run HelpingAI 15B i1?

Accepted Answer

Yes, RTX 4090 24GB can run HelpingAI 15B i1 with a C grade (Runs well). Expected decode speed: 79.3 tok/s.

Question 2

How much VRAM does HelpingAI 15B i1 need?

Accepted Answer

HelpingAI 15B i1 (15B parameters) requires approximately 15.1 GB of memory with Q4_K_M quantization.

Question 3

What is the best quantization for HelpingAI 15B i1?

Accepted Answer

The recommended quantization for HelpingAI 15B i1 is Q4_K_M, which balances quality and memory efficiency.

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	B	Runs well	79.3 tok/s	3550 ms	44K
Chat	C	Runs well	79.3 tok/s	1331 ms	14K
Coding	C	Runs well	79.3 tok/s	2441 ms	25K
RAG	B	Runs well	79.3 tok/s	4438 ms	44K
Reasoning	C	Runs well	79.3 tok/s	2885 ms	25K

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	5.9 GB	Low	D34
Q3_K_S	3	7.4 GB	Low	D35
NVFP4	4

Can RTX 4090 24GB run HelpingAI 15B i1?

Memory breakdown