Model comparison

granite-4.0-h-tiny vs Qwen3-VL-8B-Instruct-abliterated-v1

At Q4_K_M, granite-4.0-h-tiny is the smaller download — 4,230,975,936 bytes against 5,027,785,792.

From the file· summed bytes, KV per layer

Side by side

granite-4.0-h-tinyQwen3-VL-8B-Instruct-abliterated-v1
Parameters6.9B8.8B
Architecturegranitehybridqwen3vl
Layers40
Native context131,072
Mixture of expertsyes, 64 expertsno
Quantizations published3948
Smallest quantization1.81 GiB1.97 GiB
Q4_K_M3.94 GiB4.68 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextgranite-4.0-h-tinyQwen3-VL-8B-Instruct-abliterated-v1Ratio
4,0960.03 GiB
8,1920.06 GiB
16,3840.13 GiB
32,7680.25 GiB
65,5360.50 GiB
131,0721.00 GiB