Is gemma-4-E4B-it-qat-q4_0-unquantized or Qwen3.5-9B smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

gemma-4-E4B-it-qat-q4_0-unquantized — its KV cache at 32K context is 2.0x smaller, because of how its attention layers cache.

Model comparison

gemma-4-E4B-it-qat-q4_0-unquantized vs Qwen3.5-9B

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-E4B-it-qat-q4_0-unquantized's KV cache at 32K is 2.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gemma-4-E4B-it-qat-q4_0-unquantized	Qwen3.5-9B
Parameters	7.9B	9.7B
Architecture	gemma4	qwen35
Layers	42	32
Native context	131,072	262,144
Mixture of experts	no	no
Quantizations published	1	34
Smallest quantization	4.80 GiB	2.97 GiB
Q4_K_M	—	5.24 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	gemma-4-E4B-it-qat-q4_0-unquantized	Qwen3.5-9B	Ratio
4,096	0.12 GiB	0.13 GiB	1.02×
8,192	0.18 GiB	0.25 GiB	1.41×
16,384	0.29 GiB	0.50 GiB	1.74×
32,768	0.51 GiB	1.00 GiB	1.98×
65,536	0.94 GiB	2.00 GiB	2.12×
131,072	1.82 GiB	4.00 GiB	2.20×