Is WizardLM-1.0-Uncensored-Llama2-13b or Qwen3.8-27B smaller?

WizardLM-1.0-Uncensored-Llama2-13b is smaller at Q4_K_M: 7.33 GiB against 15.66 GiB.

Which uses less memory at long context?

Qwen3.8-27B — its KV cache at 32K context is 12.5x smaller, because of how its attention layers cache.

Model comparison

WizardLM-1.0-Uncensored-Llama2-13b vs Qwen3.8-27B

At Q4_K_M, WizardLM-1.0-Uncensored-Llama2-13b is the smaller download — 7,865,956,224 bytes against 16,810,714,336. At long context the gap widens: Qwen3.8-27B's KV cache at 32K is 12.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	WizardLM-1.0-Uncensored-Llama2-13b	Qwen3.8-27B
Parameters	13.0B	27.8B
Architecture	llama	qwen35
Layers	40	64
Native context	4,096	262,144
Mixture of experts	no	no
Quantizations published	12	22
Smallest quantization	5.06 GiB	8.39 GiB
Q4_K_M	7.33 GiB	15.66 GiB
Licence	llama2	apache-2.0

KV cache by context

the term that decides long-context viability

Context	WizardLM-1.0-Uncensored-Llama2-13b	Qwen3.8-27B	Ratio
4,096	3.13 GiB	0.25 GiB	12.49×
8,192	6.25 GiB	0.50 GiB	12.50×
16,384	12.50 GiB	1.00 GiB	12.50×
32,768	25.00 GiB	2.00 GiB	12.50×
65,536	50.00 GiB	4.00 GiB	12.50×
131,072	100.00 GiB	8.00 GiB	12.50×