Is llama2-22b-chat-wizard-uncensored or Qwen3.8-27B smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

Qwen3.8-27B — its KV cache at 32K context is 16.2x smaller, because of how its attention layers cache.

Model comparison

llama2-22b-chat-wizard-uncensored vs Qwen3.8-27B

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: Qwen3.8-27B's KV cache at 32K is 16.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	llama2-22b-chat-wizard-uncensored	Qwen3.8-27B
Parameters	21.8B	27.8B
Architecture	llama	qwen35
Layers	40	64
Native context	2,048	262,144
Mixture of experts	no	no
Quantizations published	2	22
Smallest quantization	7.56 GiB	8.39 GiB
Q4_K_M	—	15.66 GiB
Licence	other	apache-2.0

KV cache by context

the term that decides long-context viability

Context	llama2-22b-chat-wizard-uncensored	Qwen3.8-27B	Ratio
4,096	4.06 GiB	0.25 GiB	16.24×
8,192	8.13 GiB	0.50 GiB	16.24×
16,384	16.25 GiB	1.00 GiB	16.25×
32,768	32.50 GiB	2.00 GiB	16.25×
65,536	65.00 GiB	4.00 GiB	16.25×
131,072	130.00 GiB	8.00 GiB	16.25×