Model comparison

Qwen3.5-9B-Claude-4.6-HighIQ-INSTRUCT-HERETIC-UNCENSORED vs Qwen3-4B

At Q4_K_M, Qwen3-4B is the smaller download — 2,497,280,256 bytes against 5,627,037,568. At long context the gap widens: Qwen3.5-9B-Claude-4.6-HighIQ-INSTRUCT-HERETIC-UNCENSORED's KV cache at 32K is 4.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Qwen3.5-9B-Claude-4.6-HighIQ-INSTRUCT-HERETIC-UNCENSOREDQwen3-4B
Parameters9.4B4.0B
Architectureqwen35qwen3
Layers3236
Native context262,14440,960
Mixture of expertsnono
Quantizations published4629
Smallest quantization2.28 GiB1.01 GiB
Q4_K_M5.24 GiB2.33 GiB
Licenceapache-2.0—

KV cache by context

the term that decides long-context viability
ContextQwen3.5-9B-Claude-4.6-HighIQ-INSTRUCT-HERETIC-UNCENSOREDQwen3-4BRatio
4,0960.13 GiB0.56 GiB4.50×
8,1920.25 GiB1.13 GiB4.50×
16,3840.50 GiB2.25 GiB4.50×
32,7681.00 GiB4.50 GiB4.50×
65,5362.00 GiB9.00 GiB4.50×
131,0724.00 GiB18.00 GiB4.50×
Qwen3.5-9B-Claude-4.6-HighIQ-INSTRUCT-HERETIC-UNCENSORED vs Qwen3-4B — size, memory and hardware fit — ossmodeldb