Is SmolLM2-135M or LFM2.5-230M smaller?

SmolLM2-135M is smaller at Q4_K_M: 0.10 GiB against 0.14 GiB.

Which uses less memory at long context?

LFM2.5-230M — its KV cache at 32K context is 1.9x smaller, because of how its attention layers cache.

Model comparison

SmolLM2-135M vs LFM2.5-230M

At Q4_K_M, SmolLM2-135M is the smaller download — 105,453,568 bytes against 153,406,048. At long context the gap widens: LFM2.5-230M's KV cache at 32K is 1.9× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	SmolLM2-135M	LFM2.5-230M
Parameters	135M	230M
Architecture	llama	lfm2
Layers	30	14
Native context	8,192	128,000
Mixture of experts	no	no
Quantizations published	4	25
Smallest quantization	0.10 GiB	0.10 GiB
Q4_K_M	0.10 GiB	0.14 GiB
Licence	creativeml-openrail-m	other

KV cache by context

the term that decides long-context viability

Context	SmolLM2-135M	LFM2.5-230M	Ratio
4,096	0.09 GiB	0.05 GiB	1.87×
8,192	0.18 GiB	0.09 GiB	1.87×
16,384	0.35 GiB	0.19 GiB	1.87×
32,768	0.70 GiB	0.38 GiB	1.87×
65,536	1.41 GiB	0.75 GiB	1.87×
131,072	2.81 GiB	1.50 GiB	1.87×