Model comparison

NSFW_13B_sft vs Qwen3.6-27B

At Q4_K_M, NSFW_13B_sft is the smaller download — 8,560,487,488 bytes against 16,547,397,888. At long context the gap widens: Qwen3.6-27B's KV cache at 32K is 12.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

NSFW_13B_sftQwen3.6-27B
Parameters13.3B27.8B
Architecturebaichuanqwen35
Layers4064
Native context262,144
Mixture of expertsnono
Quantizations published1340
Smallest quantization5.18 GiB8.74 GiB
Q4_K_M7.97 GiB15.41 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextNSFW_13B_sftQwen3.6-27BRatio
4,0963.13 GiB0.25 GiB12.49×
8,1926.25 GiB0.50 GiB12.50×
16,38412.50 GiB1.00 GiB12.50×
32,76825.00 GiB2.00 GiB12.50×
65,53650.00 GiB4.00 GiB12.50×
131,072100.00 GiB8.00 GiB12.50×
NSFW_13B_sft vs Qwen3.6-27B — size, memory and hardware fit — ossmodeldb