Is Mistral-Small-3.2-24B-Instruct-2506 or gemma-4-26B-A4B-it smaller?

Mistral-Small-3.2-24B-Instruct-2506 is smaller at Q4_K_M: 13.35 GiB against 15.64 GiB.

Which uses less memory at long context?

gemma-4-26B-A4B-it — its KV cache at 32K context is 3.2x smaller, because of how its attention layers cache.

Model comparison

Mistral-Small-3.2-24B-Instruct-2506 vs gemma-4-26B-A4B-it

At Q4_K_M, Mistral-Small-3.2-24B-Instruct-2506 is the smaller download — 14,333,909,728 bytes against 16,796,017,248. At long context the gap widens: gemma-4-26B-A4B-it's KV cache at 32K is 3.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Mistral-Small-3.2-24B-Instruct-2506	gemma-4-26B-A4B-it
Parameters	24.0B	26.5B
Architecture	llama	gemma4
Layers	40	30
Native context	131,072	262,144
Mixture of experts	no	yes, 128 experts
Quantizations published	50	46
Smallest quantization	5.18 GiB	8.99 GiB
Q4_K_M	13.35 GiB	15.64 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Mistral-Small-3.2-24B-Instruct-2506	gemma-4-26B-A4B-it	Ratio
4,096	0.63 GiB	0.45 GiB	1.39×
8,192	1.25 GiB	0.61 GiB	2.06×
16,384	2.50 GiB	0.92 GiB	2.72×
32,768	5.00 GiB	1.54 GiB	3.24×
65,536	10.00 GiB	2.79 GiB	3.58×
131,072	20.00 GiB	5.29 GiB	3.78×