Is DeepSeek-V4-Flash or Hy3 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 158.8x smaller, because of how its attention layers cache.

Model comparison

DeepSeek-V4-Flash vs Hy3

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

From the file· summed bytes, KV per layer

Side by side

	DeepSeek-V4-Flash	Hy3
Parameters	291B	299B
Architecture	deepseek4	hy_v3
Layers	43	80
Native context	1,048,576	262,144
Mixture of experts	yes, 256 experts	yes, 192 experts
Quantizations published	12	37
Smallest quantization	76.87 GiB	59.65 GiB
Q4_K_M	—	168.57 GiB
Licence	mit	apache-2.0

KV cache by context

the term that decides long-context viability

Context	DeepSeek-V4-Flash	Hy3	Ratio
4,096	0.06 GiB	1.25 GiB	19.84×
8,192	0.06 GiB	2.50 GiB	39.69×
16,384	0.06 GiB	5.00 GiB	79.38×
32,768	0.06 GiB	10.00 GiB	158.76×
65,536	0.06 GiB	20.00 GiB	317.52×
131,072	0.06 GiB	40.00 GiB	635.04×