Is Qwen3.8-27B or Moonlight-16B-A3B-Instruct smaller?

Moonlight-16B-A3B-Instruct is smaller at Q4_K_M: 9.81 GiB against 15.66 GiB.

Which uses less memory at long context?

Moonlight-16B-A3B-Instruct — its KV cache at 32K context is 2.1x smaller, because of how its attention layers cache.

Model comparison

Qwen3.8-27B vs Moonlight-16B-A3B-Instruct

At Q4_K_M, Moonlight-16B-A3B-Instruct is the smaller download — 10,537,205,632 bytes against 16,810,714,336. At long context the gap widens: Moonlight-16B-A3B-Instruct's KV cache at 32K is 2.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3.8-27B	Moonlight-16B-A3B-Instruct
Parameters	27.8B	16.0B
Architecture	qwen35	deepseek2
Layers	64	27
Native context	262,144	8,192
Mixture of experts	no	yes, 64 experts
Quantizations published	22	48
Smallest quantization	8.39 GiB	4.77 GiB
Q4_K_M	15.66 GiB	9.81 GiB
Licence	apache-2.0	mit

KV cache by context

the term that decides long-context viability

Context	Qwen3.8-27B	Moonlight-16B-A3B-Instruct	Ratio
4,096	0.25 GiB	0.12 GiB	2.11×
8,192	0.50 GiB	0.24 GiB	2.11×
16,384	1.00 GiB	0.47 GiB	2.11×
32,768	2.00 GiB	0.95 GiB	2.11×
65,536	4.00 GiB	1.90 GiB	2.11×
131,072	8.00 GiB	3.80 GiB	2.11×