Is Qwen3-Zero-Coder-Reasoning-V2-0.8B or Llama-3.2-1B-Instruct smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

Llama-3.2-1B-Instruct — its KV cache at 32K context is 5.3x smaller, because of how its attention layers cache.

Model comparison

Qwen3-Zero-Coder-Reasoning-V2-0.8B vs Llama-3.2-1B-Instruct

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: Llama-3.2-1B-Instruct's KV cache at 32K is 5.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3-Zero-Coder-Reasoning-V2-0.8B	Llama-3.2-1B-Instruct
Parameters	816M	1.2B
Architecture	qwen3	llama
Layers	42	16
Native context	40,960	131,072
Mixture of experts	no	no
Quantizations published	26	39
Smallest quantization	0.24 GiB	0.39 GiB
Q4_K_M	—	0.75 GiB
Licence	apache-2.0	—

KV cache by context

the term that decides long-context viability

Context	Qwen3-Zero-Coder-Reasoning-V2-0.8B	Llama-3.2-1B-Instruct	Ratio
4,096	0.66 GiB	0.13 GiB	5.25×
8,192	1.31 GiB	0.25 GiB	5.25×
16,384	2.63 GiB	0.50 GiB	5.25×
32,768	5.25 GiB	1.00 GiB	5.25×
65,536	10.50 GiB	2.00 GiB	5.25×
131,072	21.00 GiB	4.00 GiB	5.25×