Model comparison

qwen2.5-7b-bnb-4bit vs Qwen3.6-27B

At Q4_K_M, qwen2.5-7b-bnb-4bit is the smaller download — 4,683,070,944 bytes against 16,547,397,888.

From the file· summed bytes, KV per layer

Side by side

qwen2.5-7b-bnb-4bitQwen3.6-27B
Parameters7.8B27.8B
Architectureqwen2qwen35
Layers2864
Native context131,072262,144
Mixture of expertsnono
Quantizations published640
Smallest quantization4.36 GiB8.74 GiB
Q4_K_M4.36 GiB15.41 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextqwen2.5-7b-bnb-4bitQwen3.6-27BRatio
4,0960.22 GiB0.25 GiB1.14×
8,1920.44 GiB0.50 GiB1.14×
16,3840.88 GiB1.00 GiB1.14×
32,7681.75 GiB2.00 GiB1.14×
65,5363.50 GiB4.00 GiB1.14×
131,0727.00 GiB8.00 GiB1.14×