Is Qwen3-Coder-30B-A3B-Instruct or NVIDIA-Nemotron-Nano-12B-v2 smaller?

NVIDIA-Nemotron-Nano-12B-v2 is smaller at Q4_K_M: 6.98 GiB against 17.28 GiB.

Which uses less memory at long context?

Qwen3-Coder-30B-A3B-Instruct — its KV cache at 32K context is 2.6x smaller, because of how its attention layers cache.

Model comparison

Qwen3-Coder-30B-A3B-Instruct vs NVIDIA-Nemotron-Nano-12B-v2

At Q4_K_M, NVIDIA-Nemotron-Nano-12B-v2 is the smaller download — 7,494,497,504 bytes against 18,556,689,568. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct's KV cache at 32K is 2.6× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3-Coder-30B-A3B-Instruct	NVIDIA-Nemotron-Nano-12B-v2
Parameters	30.5B	12.3B
Architecture	qwen3moe	nemotron_h
Layers	48	62
Native context	262,144	131,072
Mixture of experts	yes, 128 experts	no
Quantizations published	46	30
Smallest quantization	7.46 GiB	3.79 GiB
Q4_K_M	17.28 GiB	6.98 GiB
Licence	apache-2.0	—

KV cache by context

the term that decides long-context viability

Context	Qwen3-Coder-30B-A3B-Instruct	NVIDIA-Nemotron-Nano-12B-v2	Ratio
4,096	0.38 GiB	0.97 GiB	2.58×
8,192	0.75 GiB	1.94 GiB	2.58×
16,384	1.50 GiB	3.88 GiB	2.58×
32,768	3.00 GiB	7.75 GiB	2.58×
65,536	6.00 GiB	15.50 GiB	2.58×
131,072	12.00 GiB	31.00 GiB	2.58×