Is bitnet-b1.58-2B-4T or gemma-3-270m-it smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

gemma-3-270m-it — its KV cache at 32K context is 21.6x smaller, because of how its attention layers cache.

Model comparison

bitnet-b1.58-2B-4T vs gemma-3-270m-it

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-3-270m-it's KV cache at 32K is 21.6× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	bitnet-b1.58-2B-4T	gemma-3-270m-it
Parameters	850M	268M
Architecture	—	gemma3
Layers	30	18
Native context	4,096	32,768
Mixture of experts	no	no
Quantizations published	2	39
Smallest quantization	0.10 GiB	0.17 GiB
Q4_K_M	—	0.24 GiB
Licence	mit	gemma

KV cache by context

the term that decides long-context viability

Context	bitnet-b1.58-2B-4T	gemma-3-270m-it	Ratio
4,096	0.29 GiB	0.03 GiB	11.11×
8,192	0.59 GiB	0.04 GiB	15.38×
16,384	1.17 GiB	0.06 GiB	19.05×
32,768	2.34 GiB	0.11 GiB	21.62×
65,536	4.69 GiB	0.20 GiB	23.19×
131,072	9.38 GiB	0.39 GiB	24.06×