Is Llama-3.2-3B-Instruct-abliterated or Qwen3-8B smaller?

Llama-3.2-3B-Instruct-abliterated is smaller at Q4_K_M: 2.09 GiB against 4.68 GiB.

Which uses less memory at long context?

Llama-3.2-3B-Instruct-abliterated — its KV cache at 32K context is 1.3x smaller, because of how its attention layers cache.

Model comparison

Llama-3.2-3B-Instruct-abliterated vs Qwen3-8B

At Q4_K_M, Llama-3.2-3B-Instruct-abliterated is the smaller download — 2,241,004,000 bytes against 5,027,783,488. At long context the gap widens: Llama-3.2-3B-Instruct-abliterated's KV cache at 32K is 1.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Llama-3.2-3B-Instruct-abliterated	Qwen3-8B
Parameters	3.6B	8.2B
Architecture	llama	qwen3
Layers	28	36
Native context	131,072	40,960
Mixture of experts	no	no
Quantizations published	37	51
Smallest quantization	0.93 GiB	2.12 GiB
Q4_K_M	2.09 GiB	4.68 GiB
Licence	—	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Llama-3.2-3B-Instruct-abliterated	Qwen3-8B	Ratio
4,096	0.44 GiB	0.56 GiB	1.29×
8,192	0.88 GiB	1.13 GiB	1.29×
16,384	1.75 GiB	2.25 GiB	1.29×
32,768	3.50 GiB	4.50 GiB	1.29×
65,536	7.00 GiB	9.00 GiB	1.29×
131,072	14.00 GiB	18.00 GiB	1.29×