Is Qwen3.5-9B or Qwen3-VL-32B-Instruct-ultra-uncensored-heretic smaller?

Qwen3.5-9B is smaller at Q4_K_M: 5.24 GiB against 18.40 GiB.

Which uses less memory at long context?

Qwen3.5-9B — its KV cache at 32K context is 8.0x smaller, because of how its attention layers cache.

Model comparison

Qwen3.5-9B vs Qwen3-VL-32B-Instruct-ultra-uncensored-heretic

At Q4_K_M, Qwen3.5-9B is the smaller download — 5,627,044,256 bytes against 19,762,151,392. At long context the gap widens: Qwen3.5-9B's KV cache at 32K is 8.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3.5-9B	Qwen3-VL-32B-Instruct-ultra-uncensored-heretic
Parameters	9.7B	33.4B
Architecture	qwen35	qwen3vl
Layers	32	64
Native context	262,144	262,144
Mixture of experts	no	no
Quantizations published	34	43
Smallest quantization	2.97 GiB	6.82 GiB
Q4_K_M	5.24 GiB	18.40 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Qwen3.5-9B	Qwen3-VL-32B-Instruct-ultra-uncensored-heretic	Ratio
4,096	0.13 GiB	1.00 GiB	7.99×
8,192	0.25 GiB	2.00 GiB	8.00×
16,384	0.50 GiB	4.00 GiB	8.00×
32,768	1.00 GiB	8.00 GiB	8.00×
65,536	2.00 GiB	16.00 GiB	8.00×
131,072	4.00 GiB	32.00 GiB	8.00×