Model comparison

Dolphin3.0-Qwen2.5-3b vs Qwen3-VL-8B-Instruct-abliterated-v1

At Q4_K_M, Dolphin3.0-Qwen2.5-3b is the smaller download — 1,929,905,856 bytes against 5,027,785,792.

From the file· summed bytes, KV per layer

Side by side

Dolphin3.0-Qwen2.5-3bQwen3-VL-8B-Instruct-abliterated-v1
Parameters3.1B8.8B
Architectureqwen2qwen3vl
Layers36
Native context32,768
Mixture of expertsnono
Quantizations published4748
Smallest quantization1.06 GiB1.97 GiB
Q4_K_M1.80 GiB4.68 GiB
Licenceotherapache-2.0

KV cache by context

the term that decides long-context viability
ContextDolphin3.0-Qwen2.5-3bQwen3-VL-8B-Instruct-abliterated-v1Ratio
4,0960.14 GiB
8,1920.28 GiB
16,3840.56 GiB
32,7681.13 GiB
65,5362.25 GiB
131,0724.50 GiB