Is DeepSeek-V4-Flash or Trinity-Large-TrueBase 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 42.3x smaller, because of how its attention layers cache.

Model comparison

DeepSeek-V4-Flash vs Trinity-Large-TrueBase

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

From the file· summed bytes, KV per layer

Side by side

	DeepSeek-V4-Flash	Trinity-Large-TrueBase
Parameters	291B	399B
Architecture	deepseek4	afmoe
Layers	43	60
Native context	1,048,576	8,192
Mixture of experts	yes, 256 experts	yes, 256 experts
Quantizations published	12	39
Smallest quantization	76.87 GiB	73.49 GiB
Q4_K_M	—	225.04 GiB
Licence	mit	other

KV cache by context

the term that decides long-context viability

Context	DeepSeek-V4-Flash	Trinity-Large-TrueBase	Ratio
4,096	0.06 GiB	0.94 GiB	14.88×
8,192	0.06 GiB	1.26 GiB	20.00×
16,384	0.06 GiB	1.73 GiB	27.44×
32,768	0.06 GiB	2.67 GiB	42.33×
65,536	0.06 GiB	4.54 GiB	72.09×
131,072	0.06 GiB	8.29 GiB	131.63×