Is Solar-Open2-250B 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 95.3x smaller, because of how its attention layers cache.

Model comparison

Solar-Open2-250B 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 95.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Solar-Open2-250B	DeepSeek-V4-Flash
Parameters	250B	291B
Architecture	solar-open2	deepseek4
Layers	48	43
Native context	1,048,576	1,048,576
Mixture of experts	yes, 320 experts	yes, 256 experts
Quantizations published	13	12
Smallest quantization	52.67 GiB	76.87 GiB
Q4_K_M	141.39 GiB	—
Licence	other	mit

KV cache by context

the term that decides long-context viability

Context	Solar-Open2-250B	DeepSeek-V4-Flash	Ratio
4,096	0.75 GiB	0.06 GiB	11.91×
8,192	1.50 GiB	0.06 GiB	23.81×
16,384	3.00 GiB	0.06 GiB	47.63×
32,768	6.00 GiB	0.06 GiB	95.26×
65,536	12.00 GiB	0.06 GiB	190.51×
131,072	24.00 GiB	0.06 GiB	381.02×