Model comparison

Vikhr-Nemo-12B-Instruct-R-21-09-24 vs Qwen3.6-27B

At Q4_K_M, Vikhr-Nemo-12B-Instruct-R-21-09-24 is the smaller download — 7,477,218,656 bytes against 16,547,397,888. At long context the gap widens: Qwen3.6-27B's KV cache at 32K is 2.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Vikhr-Nemo-12B-Instruct-R-21-09-24Qwen3.6-27B
Parameters12.2B27.8B
Architecturellamaqwen35
Layers4064
Native context1,024,000262,144
Mixture of expertsnono
Quantizations published2240
Smallest quantization4.13 GiB8.74 GiB
Q4_K_M6.96 GiB15.41 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextVikhr-Nemo-12B-Instruct-R-21-09-24Qwen3.6-27BRatio
4,0960.63 GiB0.25 GiB2.50×
8,1921.25 GiB0.50 GiB2.50×
16,3842.50 GiB1.00 GiB2.50×
32,7685.00 GiB2.00 GiB2.50×
65,53610.00 GiB4.00 GiB2.50×
131,07220.00 GiB8.00 GiB2.50×