Model comparison

Qwen3.6-27B vs Seed-OSS-36B-Instruct-biprojected-norm-preserving-abliterated

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: Qwen3.6-27B's KV cache at 32K is 4.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Qwen3.6-27BSeed-OSS-36B-Instruct-biprojected-norm-preserving-abliterated
Parameters27.8B36.2B
Architectureqwen35seed_oss
Layers6464
Native context262,144524,288
Mixture of expertsnono
Quantizations published4046
Smallest quantization8.74 GiB7.50 GiB
Q4_K_M15.41 GiB
Licenceapache-2.0

KV cache by context

the term that decides long-context viability
ContextQwen3.6-27BSeed-OSS-36B-Instruct-biprojected-norm-preserving-abliteratedRatio
4,0960.25 GiB1.00 GiB4.00×
8,1920.50 GiB2.00 GiB4.00×
16,3841.00 GiB4.00 GiB4.00×
32,7682.00 GiB8.00 GiB4.00×
65,5364.00 GiB16.00 GiB4.00×
131,0728.00 GiB32.00 GiB4.00×