Is Vikhr-Gemma-2B-instruct or Qwen3-1.7B smaller?

Qwen3-1.7B is smaller at Q4_K_M: 1.03 GiB against 1.59 GiB.

Which uses less memory at long context?

Vikhr-Gemma-2B-instruct — its KV cache at 32K context is 1.9x smaller, because of how its attention layers cache.

Model comparison

Vikhr-Gemma-2B-instruct vs Qwen3-1.7B

At Q4_K_M, Qwen3-1.7B is the smaller download — 1,107,409,472 bytes against 1,708,582,848. At long context the gap widens: Vikhr-Gemma-2B-instruct'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

	Vikhr-Gemma-2B-instruct	Qwen3-1.7B
Parameters	2.6B	2.0B
Architecture	gemma2	qwen3
Layers	26	28
Native context	8,192	40,960
Mixture of experts	no	no
Quantizations published	46	48
Smallest quantization	0.78 GiB	0.50 GiB
Q4_K_M	1.59 GiB	1.03 GiB
Licence	apache-2.0	—

KV cache by context

the term that decides long-context viability

Context	Vikhr-Gemma-2B-instruct	Qwen3-1.7B	Ratio
4,096	0.41 GiB	0.44 GiB	1.08×
8,192	0.63 GiB	0.88 GiB	1.38×
16,384	1.04 GiB	1.75 GiB	1.68×
32,768	1.85 GiB	3.50 GiB	1.89×
65,536	3.48 GiB	7.00 GiB	2.01×
131,072	6.73 GiB	14.00 GiB	2.08×