Is deepseek-math-7b-instruct or gemma-4-E4B-it smaller?

deepseek-math-7b-instruct is smaller at Q4_K_M: 3.93 GiB against 4.64 GiB.

Which uses less memory at long context?

gemma-4-E4B-it — its KV cache at 32K context is 29.7x smaller, because of how its attention layers cache.

Model comparison

deepseek-math-7b-instruct vs gemma-4-E4B-it

At Q4_K_M, deepseek-math-7b-instruct is the smaller download — 4,221,964,704 bytes against 4,977,171,584. At long context the gap widens: gemma-4-E4B-it's KV cache at 32K is 29.7× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	deepseek-math-7b-instruct	gemma-4-E4B-it
Parameters	6.9B	8.0B
Architecture	llama	gemma4
Layers	30	42
Native context	4,096	131,072
Mixture of experts	no	no
Quantizations published	38	36
Smallest quantization	2.53 GiB	3.30 GiB
Q4_K_M	3.93 GiB	4.64 GiB
Licence	other	apache-2.0

KV cache by context

the term that decides long-context viability

Context	deepseek-math-7b-instruct	gemma-4-E4B-it	Ratio
4,096	1.88 GiB	0.12 GiB	15.24×
8,192	3.75 GiB	0.18 GiB	21.10×
16,384	7.50 GiB	0.29 GiB	26.12×
32,768	15.00 GiB	0.51 GiB	29.65×
65,536	30.00 GiB	0.94 GiB	31.80×
131,072	60.00 GiB	1.82 GiB	33.00×