Is Qwen3.5-4B or Apriel-1.6-15b-Thinker smaller?

Qwen3.5-4B is smaller at Q4_K_M: 2.52 GiB against 8.18 GiB.

Which uses less memory at long context?

Qwen3.5-4B — its KV cache at 32K context is 6.0x smaller, because of how its attention layers cache.

Model comparison

Qwen3.5-4B vs Apriel-1.6-15b-Thinker

At Q4_K_M, Qwen3.5-4B is the smaller download — 2,707,513,696 bytes against 8,785,478,144. At long context the gap widens: Qwen3.5-4B's KV cache at 32K is 6.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3.5-4B	Apriel-1.6-15b-Thinker
Parameters	4.7B	14.9B
Architecture	qwen35	llama
Layers	32	48
Native context	262,144	262,400
Mixture of experts	no	no
Quantizations published	57	60
Smallest quantization	1.42 GiB	3.22 GiB
Q4_K_M	2.52 GiB	8.18 GiB
Licence	apache-2.0	mit

KV cache by context

the term that decides long-context viability

Context	Qwen3.5-4B	Apriel-1.6-15b-Thinker	Ratio
4,096	0.13 GiB	0.75 GiB	6.00×
8,192	0.25 GiB	1.50 GiB	6.00×
16,384	0.50 GiB	3.00 GiB	6.00×
32,768	1.00 GiB	6.00 GiB	6.00×
65,536	2.00 GiB	12.00 GiB	6.00×
131,072	4.00 GiB	24.00 GiB	6.00×