Model comparison

OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored vs Qwen3.6-27B

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored's KV cache at 32K is 2.6× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-UncensoredQwen3.6-27B
Parameters20.9B27.8B
Architecturegpt-ossqwen35
Layers2464
Native context131,072262,144
Mixture of expertsyes, 32 expertsno
Quantizations published4740
Smallest quantization11.19 GiB8.74 GiB
Q4_K_M15.41 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextOpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-UncensoredQwen3.6-27BRatio
4,0960.11 GiB0.25 GiB2.25×
8,1920.21 GiB0.50 GiB2.44×
16,3840.39 GiB1.00 GiB2.55×
32,7680.77 GiB2.00 GiB2.61×
65,5361.52 GiB4.00 GiB2.64×
131,0723.02 GiB8.00 GiB2.65×