Model comparison

Jan-v2-VL-high vs Qwythos-9B-Claude-Mythos-5-1M

At Q4_K_M, Jan-v2-VL-high is the smaller download — 5,027,784,672 bytes against 11,516,776,704. At long context the gap widens: Qwythos-9B-Claude-Mythos-5-1M'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

Jan-v2-VL-highQwythos-9B-Claude-Mythos-5-1M
Parameters8.8B9.4B
Architectureqwen3vlqwen35
Layers3632
Native context262,1441,048,576
Mixture of expertsnono
Quantizations published2516
Smallest quantization3.06 GiB5.38 GiB
Q4_K_M4.68 GiB10.73 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextJan-v2-VL-highQwythos-9B-Claude-Mythos-5-1MRatio
4,0960.56 GiB0.13 GiB4.50×
8,1921.13 GiB0.25 GiB4.50×
16,3842.25 GiB0.50 GiB4.50×
32,7684.50 GiB1.00 GiB4.50×
65,5369.00 GiB2.00 GiB4.50×
131,07218.00 GiB4.00 GiB4.50×