Model comparison

TinyLlama-1.1B-Chat-v1.0 vs Qwen3-1.7B

At Q4_K_M, TinyLlama-1.1B-Chat-v1.0 is the smaller download — 667,814,880 bytes against 1,107,409,472. At long context the gap widens: TinyLlama-1.1B-Chat-v1.0's KV cache at 32K is 5.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

TinyLlama-1.1B-Chat-v1.0Qwen3-1.7B
Parameters1.1B2.0B
Architecturellamaqwen3
Layers2228
Native context2,04840,960
Mixture of expertsnono
Quantizations published2648
Smallest quantization0.45 GiB0.50 GiB
Q4_K_M0.62 GiB1.03 GiB
Licenceapache-2.0

KV cache by context

the term that decides long-context viability
ContextTinyLlama-1.1B-Chat-v1.0Qwen3-1.7BRatio
4,0960.09 GiB0.44 GiB5.09×
8,1920.17 GiB0.88 GiB5.09×
16,3840.34 GiB1.75 GiB5.09×
32,7680.69 GiB3.50 GiB5.09×
65,5361.38 GiB7.00 GiB5.09×
131,0722.75 GiB14.00 GiB5.09×