Is Llama-3.2-1B-Instruct-abliterated or Qwen3-4B smaller?

Llama-3.2-1B-Instruct-abliterated is smaller at Q4_K_M: 0.89 GiB against 2.33 GiB.

Which uses less memory at long context?

Llama-3.2-1B-Instruct-abliterated — its KV cache at 32K context is 4.5x smaller, because of how its attention layers cache.

Model comparison

Llama-3.2-1B-Instruct-abliterated vs Qwen3-4B

At Q4_K_M, Llama-3.2-1B-Instruct-abliterated is the smaller download — 955,445,792 bytes against 2,497,280,256. At long context the gap widens: Llama-3.2-1B-Instruct-abliterated's KV cache at 32K is 4.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Llama-3.2-1B-Instruct-abliterated	Qwen3-4B
Parameters	1.5B	4.0B
Architecture	llama	qwen3
Layers	16	36
Native context	131,072	40,960
Mixture of experts	no	no
Quantizations published	38	29
Smallest quantization	0.45 GiB	1.01 GiB
Q4_K_M	0.89 GiB	2.33 GiB
Licence	llama3.2	—

KV cache by context

the term that decides long-context viability

Context	Llama-3.2-1B-Instruct-abliterated	Qwen3-4B	Ratio
4,096	0.13 GiB	0.56 GiB	4.50×
8,192	0.25 GiB	1.13 GiB	4.50×
16,384	0.50 GiB	2.25 GiB	4.50×
32,768	1.00 GiB	4.50 GiB	4.50×
65,536	2.00 GiB	9.00 GiB	4.50×
131,072	4.00 GiB	18.00 GiB	4.50×