Is MiniMax-M2.7 or DeepSeek-V4-Flash smaller?

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

Which uses less memory at long context?

DeepSeek-V4-Flash — its KV cache at 32K context is 123.0x smaller, because of how its attention layers cache.

Model comparison

MiniMax-M2.7 vs DeepSeek-V4-Flash

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: DeepSeek-V4-Flash's KV cache at 32K is 123.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	MiniMax-M2.7	DeepSeek-V4-Flash
Parameters	229B	291B
Architecture	minimax-m2	deepseek4
Layers	62	43
Native context	204,800	1,048,576
Mixture of experts	yes, 256 experts	yes, 256 experts
Quantizations published	44	12
Smallest quantization	43.47 GiB	76.87 GiB
Q4_K_M	128.84 GiB	—
Licence	other	mit

KV cache by context

the term that decides long-context viability

Context	MiniMax-M2.7	DeepSeek-V4-Flash	Ratio
4,096	0.97 GiB	0.06 GiB	15.38×
8,192	1.94 GiB	0.06 GiB	30.76×
16,384	3.88 GiB	0.06 GiB	61.52×
32,768	7.75 GiB	0.06 GiB	123.04×
65,536	15.50 GiB	0.06 GiB	246.08×
131,072	31.00 GiB	0.06 GiB	492.16×