Is Ministral-8B-Instruct-2410 or gemma-4-12B-it-qat-q4_0-unquantized 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 1.8x smaller, because of how its attention layers cache.

Model comparison

Ministral-8B-Instruct-2410 vs gemma-4-12B-it-qat-q4_0-unquantized

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 1.8× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Ministral-8B-Instruct-2410	gemma-4-12B-it-qat-q4_0-unquantized
Parameters	8.0B	12.0B
Architecture	llama	gemma4
Layers	36	48
Native context	32,768	262,144
Mixture of experts	no	no
Quantizations published	21	2
Smallest quantization	2.76 GiB	6.50 GiB
Q4_K_M	4.57 GiB	—
Licence	other	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Ministral-8B-Instruct-2410	gemma-4-12B-it-qat-q4_0-unquantized	Ratio
4,096	0.56 GiB	0.72 GiB	1.28×
8,192	1.13 GiB	0.97 GiB	1.16×
16,384	2.25 GiB	1.47 GiB	1.53×
32,768	4.50 GiB	2.47 GiB	1.82×
65,536	5.68 GiB	4.47 GiB	1.27×
131,072	7.93 GiB	8.47 GiB	1.07×