Is gpt-oss-120b 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 18.3x smaller, because of how its attention layers cache.

Model comparison

gpt-oss-120b 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 18.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gpt-oss-120b	DeepSeek-V4-Flash
Parameters	120B	291B
Architecture	gpt-oss	deepseek4
Layers	36	43
Native context	131,072	1,048,576
Mixture of experts	yes, 128 experts	yes, 256 experts
Quantizations published	15	12
Smallest quantization	58.27 GiB	76.87 GiB
Q4_K_M	58.46 GiB	—
Licence	apache-2.0	mit

KV cache by context

the term that decides long-context viability

Context	gpt-oss-120b	DeepSeek-V4-Flash	Ratio
4,096	0.17 GiB	0.06 GiB	2.65×
8,192	0.31 GiB	0.06 GiB	4.88×
16,384	0.59 GiB	0.06 GiB	9.35×
32,768	1.15 GiB	0.06 GiB	18.28×
65,536	2.28 GiB	0.06 GiB	36.14×
131,072	4.53 GiB	0.06 GiB	71.86×