Model comparison

Qwen3.8-27B vs LocalAI-functioncall-qwen2.5-7b-v0.5

At Q4_K_M, LocalAI-functioncall-qwen2.5-7b-v0.5 is the smaller download — 4,683,074,048 bytes against 16,810,714,336.

From the file· summed bytes, KV per layer

Side by side

Qwen3.8-27BLocalAI-functioncall-qwen2.5-7b-v0.5
Parameters27.8B7.6B
Architectureqwen35qwen2
Layers6428
Native context262,14432,768
Mixture of expertsnono
Quantizations published221
Smallest quantization8.39 GiB4.36 GiB
Q4_K_M15.66 GiB4.36 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextQwen3.8-27BLocalAI-functioncall-qwen2.5-7b-v0.5Ratio
4,0960.25 GiB0.22 GiB1.14×
8,1920.50 GiB0.44 GiB1.14×
16,3841.00 GiB0.88 GiB1.14×
32,7682.00 GiB1.75 GiB1.14×
65,5364.00 GiB3.50 GiB1.14×
131,0728.00 GiB7.00 GiB1.14×