Is step-3.5-flash 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 206.9x smaller, because of how its attention layers cache.

Model comparison

step-3.5-flash 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 206.9× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	step-3.5-flash	DeepSeek-V4-Flash
Parameters	199B	291B
Architecture	step35	deepseek4
Layers	45	43
Native context	262,144	1,048,576
Mixture of experts	no	yes, 256 experts
Quantizations published	30	12
Smallest quantization	38.47 GiB	76.87 GiB
Q4_K_M	111.70 GiB	—
Licence	apache-2.0	mit

KV cache by context

the term that decides long-context viability

Context	step-3.5-flash	DeepSeek-V4-Flash	Ratio
4,096	2.53 GiB	0.06 GiB	40.19×
8,192	4.03 GiB	0.06 GiB	64.00×
16,384	7.03 GiB	0.06 GiB	111.63×
32,768	13.03 GiB	0.06 GiB	206.88×
65,536	25.03 GiB	0.06 GiB	397.40×
131,072	49.03 GiB	0.06 GiB	778.42×