Is bitnet-b1.58-2B-4T or Qwen3-0.6B smaller?

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

Which uses less memory at long context?

bitnet-b1.58-2B-4T — its KV cache at 32K context is 1.5x smaller, because of how its attention layers cache.

Model comparison

bitnet-b1.58-2B-4T vs Qwen3-0.6B

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: bitnet-b1.58-2B-4T's KV cache at 32K is 1.5× 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	Qwen3-0.6B
Parameters	850M	752M
Architecture	—	qwen3
Layers	30	28
Native context	4,096	40,960
Mixture of experts	no	no
Quantizations published	2	44
Smallest quantization	0.10 GiB	0.20 GiB
Q4_K_M	—	0.37 GiB
Licence	mit	—

KV cache by context

the term that decides long-context viability

Context	bitnet-b1.58-2B-4T	Qwen3-0.6B	Ratio
4,096	0.29 GiB	0.44 GiB	1.49×
8,192	0.59 GiB	0.88 GiB	1.49×
16,384	1.17 GiB	1.75 GiB	1.49×
32,768	2.34 GiB	3.50 GiB	1.49×
65,536	4.69 GiB	7.00 GiB	1.49×
131,072	9.38 GiB	14.00 GiB	1.49×