Is Qwen3-Coder-30B-A3B-Instruct or gemma-4-A4B-98e-v6-coder-it smaller?

Qwen3-Coder-30B-A3B-Instruct is smaller at Q4_K_M: 17.28 GiB against 22.25 GiB.

Which uses less memory at long context?

gemma-4-A4B-98e-v6-coder-it — its KV cache at 32K context is 1.9x smaller, because of how its attention layers cache.

Model comparison

Qwen3-Coder-30B-A3B-Instruct vs gemma-4-A4B-98e-v6-coder-it

At Q4_K_M, Qwen3-Coder-30B-A3B-Instruct is the smaller download — 18,556,689,568 bytes against 23,887,604,320. At long context the gap widens: gemma-4-A4B-98e-v6-coder-it's KV cache at 32K is 1.9× 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	gemma-4-A4B-98e-v6-coder-it
Parameters	30.5B	20.5B
Architecture	qwen3moe	gemma4
Layers	48	30
Native context	262,144	262,144
Mixture of experts	yes, 128 experts	yes, 98 experts
Quantizations published	46	23
Smallest quantization	7.46 GiB	7.24 GiB
Q4_K_M	17.28 GiB	22.25 GiB
Licence	apache-2.0	gemma

KV cache by context

the term that decides long-context viability

Context	Qwen3-Coder-30B-A3B-Instruct	gemma-4-A4B-98e-v6-coder-it	Ratio
4,096	0.38 GiB	0.45 GiB	1.20×
8,192	0.75 GiB	0.61 GiB	1.24×
16,384	1.50 GiB	0.92 GiB	1.63×
32,768	3.00 GiB	1.54 GiB	1.94×
65,536	6.00 GiB	2.79 GiB	2.15×
131,072	12.00 GiB	5.29 GiB	2.27×