How much VRAM does polyglot-lion-1.7b need?

Q4_K_M is exactly 1,107,408,160 bytes (1.03 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

knoveleng · text

polyglot-lion-1.7b

knoveleng/polyglot-lion-1.7b

polyglot-lion-1.7b at Q4_K_M is exactly 1,107,408,160 bytes (1.03 GiB / 1.11 GB) — an effective 4.347 bits per weight, not the nominal 4.

From the file· summed from 1 file(s)From the file· KV per layer

Parameters

2.0B

Architecture

qwen3vl

null layers

Context

—

native (config.json)

License

mit

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
I1-IQ1_S	0.48 GiB	515,776,032	2.025	—	mradermacher
I1-IQ1_M	0.51 GiB	543,792,672	2.135	—	mradermacher
I1-IQ2_XXS	0.55 GiB	590,487,072	2.318	—	mradermacher
I1-IQ2_XS	0.59 GiB	631,512,608	2.479	—	mradermacher
I1-IQ2_S	0.61 GiB	657,825,312	2.582	—	mradermacher
I1-IQ2_M	0.65 GiB	695,180,832	2.729	—	mradermacher
I1-Q2_K_S	0.68 GiB	732,968,480	2.877	—	mradermacher
I1-IQ3_XXS	0.70 GiB	754,359,840	2.961	—	mradermacher
Q2_K	0.72 GiB	777,794,848	3.053	—	mradermacher
I1-Q2_K	0.72 GiB	777,795,104	3.053	—	mradermacher
I1-IQ3_XS	0.78 GiB	834,221,600	3.275	—	mradermacher
Q3_K_S	0.81 GiB	867,251,488	3.404	—	mradermacher
I1-IQ3_S	0.81 GiB	867,251,744	3.404	—	mradermacher
I1-Q3_K_S	0.81 GiB	867,251,744	3.404	—	mradermacher
I1-IQ3_M	0.83 GiB	895,661,600	3.516	—	mradermacher
Q3_K_M	0.88 GiB	939,537,696	3.688	—	mradermacher
I1-Q3_K_M	0.88 GiB	939,537,952	3.688	—	mradermacher
Q3_K_L	0.93 GiB	1,003,500,832	3.939	—	mradermacher
I1-Q3_K_L	0.93 GiB	1,003,501,088	3.939	—	mradermacher
I1-IQ4_XS	0.94 GiB	1,010,382,368	3.966	—	mradermacher
IQ4_XS	0.95 GiB	1,016,280,352	3.989	—	mradermacher
I1-IQ4_NL	0.98 GiB	1,054,422,560	4.139	—	mradermacher
I1-Q4_0	0.98 GiB	1,056,781,856	4.148	—	mradermacher
Q4_K_S	0.99 GiB	1,060,189,472	4.162	—	mradermacher
I1-Q4_K_S	0.99 GiB	1,060,189,728	4.162	—	mradermacher
Q4_K_M	1.03 GiB	1,107,408,160	4.347	—	mradermacher
I1-Q4_K_M	1.03 GiB	1,107,408,416	4.347	—	mradermacher
I1-Q4_1	1.06 GiB	1,142,502,944	4.485	—	mradermacher
Q5_K_S	1.15 GiB	1,230,583,072	4.830	—	mradermacher
I1-Q5_K_S	1.15 GiB	1,230,583,328	4.830	—	mradermacher
Q5_K_M	1.17 GiB	1,257,878,816	4.938	—	mradermacher
I1-Q5_K_M	1.17 GiB	1,257,879,072	4.938	—	mradermacher
Q6_K	1.32 GiB	1,417,753,888	5.565	—	mradermacher
I1-Q6_K	1.32 GiB	1,417,754,144	5.565	—	mradermacher
Q8_0	1.71 GiB	1,834,425,632	7.201	—	mradermacher
F16	3.21 GiB	3,447,348,512	13.532	—	mradermacher

Compare with

same modality, comparable size

vs Qwen3-4B vs HyperCLOVAX-SEED-Text-Instruct-1.5B vs Llama-3.2-1B-Instruct vs ced-base vs gemma-3-1b-it vs Qwen3-1.7B All alternatives →Will it fit on my card? →

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 1.07 GiB. The real file is 1.03 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json

Layers

—

Attention heads

—

KV heads

—

Head dim

—

Hidden size

—

Vocab

—

Sliding window

none

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does polyglot-lion-1.7b need?: Q4_K_M is exactly 1,107,408,160 bytes (1.03 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of polyglot-lion-1.7b should I use?: Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.