Is gemma-4-12B-it-qat-q4_0-unquantized or deepseek-coder-33b-instruct smaller?

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

Which uses less memory at long context?

gemma-4-12B-it-qat-q4_0-unquantized — its KV cache at 32K context is 3.1x smaller, because of how its attention layers cache.

Model comparison

gemma-4-12B-it-qat-q4_0-unquantized vs deepseek-coder-33b-instruct

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-12B-it-qat-q4_0-unquantized's KV cache at 32K is 3.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gemma-4-12B-it-qat-q4_0-unquantized	deepseek-coder-33b-instruct
Parameters	12.0B	33.3B
Architecture	gemma4	llama
Layers	48	62
Native context	262,144	16,384
Mixture of experts	no	no
Quantizations published	2	26
Smallest quantization	6.50 GiB	11.51 GiB
Q4_K_M	—	18.57 GiB
Licence	apache-2.0	other

KV cache by context

the term that decides long-context viability

Context	gemma-4-12B-it-qat-q4_0-unquantized	deepseek-coder-33b-instruct	Ratio
4,096	0.72 GiB	0.97 GiB	1.35×
8,192	0.97 GiB	1.94 GiB	2.00×
16,384	1.47 GiB	3.88 GiB	2.64×
32,768	2.47 GiB	7.75 GiB	3.14×
65,536	4.47 GiB	15.50 GiB	3.47×
131,072	8.47 GiB	31.00 GiB	3.66×