Is next-1b or Llama-3.2-1B-Instruct smaller?

next-1b is smaller at Q4_K_M: 0.75 GiB against 0.75 GiB.

Which uses less memory at long context?

next-1b — its KV cache at 32K context is 6.8x smaller, because of how its attention layers cache.

Model comparison

next-1b vs Llama-3.2-1B-Instruct

At Q4_K_M, next-1b is the smaller download — 806,057,664 bytes against 807,690,688. At long context the gap widens: next-1b's KV cache at 32K is 6.8× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	next-1b	Llama-3.2-1B-Instruct
Parameters	1000M	1.2B
Architecture	gemma3	llama
Layers	26	16
Native context	32,768	131,072
Mixture of experts	no	no
Quantizations published	1	39
Smallest quantization	0.75 GiB	0.39 GiB
Q4_K_M	0.75 GiB	0.75 GiB
Licence	mit	—

KV cache by context

the term that decides long-context viability

Context	next-1b	Llama-3.2-1B-Instruct	Ratio
4,096	0.04 GiB	0.13 GiB	3.37×
8,192	0.05 GiB	0.25 GiB	4.74×
16,384	0.08 GiB	0.50 GiB	5.95×
32,768	0.15 GiB	1.00 GiB	6.83×
65,536	0.27 GiB	2.00 GiB	7.37×
131,072	0.52 GiB	4.00 GiB	7.67×