Model comparison

Semancer-12B vs Qwen3-VL-8B-Instruct-abliterated-v1

At Q4_K_M, Qwen3-VL-8B-Instruct-abliterated-v1 is the smaller download — 5,027,785,792 bytes against 7,381,382,496.

From the file· summed bytes, KV per layer

Side by side

Semancer-12BQwen3-VL-8B-Instruct-abliterated-v1
Parameters12.0B8.8B
Architecturegemma4qwen3vl
Layers48
Native context262,144
Mixture of expertsnono
Quantizations published3548
Smallest quantization2.78 GiB1.97 GiB
Q4_K_M6.87 GiB4.68 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextSemancer-12BQwen3-VL-8B-Instruct-abliterated-v1Ratio
4,0960.72 GiB
8,1920.97 GiB
16,3841.47 GiB
32,7682.47 GiB
65,5364.47 GiB
131,0728.47 GiB
Semancer-12B vs Qwen3-VL-8B-Instruct-abliterated-v1 — size, memory and hardware fit — ossmodeldb