Is Qwen3-Coder-30B-A3B-Instruct or llama-3-firefunction-v2 smaller?

Qwen3-Coder-30B-A3B-Instruct is smaller at Q4_K_M: 17.28 GiB against 39.60 GiB.

Which uses less memory at long context?

Qwen3-Coder-30B-A3B-Instruct — its KV cache at 32K context is 3.3x smaller, because of how its attention layers cache.

Model comparison

Qwen3-Coder-30B-A3B-Instruct vs llama-3-firefunction-v2

At Q4_K_M, Qwen3-Coder-30B-A3B-Instruct is the smaller download — 18,556,689,568 bytes against 42,520,396,416. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct's KV cache at 32K is 3.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3-Coder-30B-A3B-Instruct	llama-3-firefunction-v2
Parameters	30.5B	70.6B
Architecture	qwen3moe	llama
Layers	48	80
Native context	262,144	8,192
Mixture of experts	yes, 128 experts	no
Quantizations published	46	15
Smallest quantization	7.46 GiB	14.29 GiB
Q4_K_M	17.28 GiB	39.60 GiB
Licence	apache-2.0	llama3

KV cache by context

the term that decides long-context viability

Context	Qwen3-Coder-30B-A3B-Instruct	llama-3-firefunction-v2	Ratio
4,096	0.38 GiB	1.25 GiB	3.33×
8,192	0.75 GiB	2.50 GiB	3.33×
16,384	1.50 GiB	5.00 GiB	3.33×
32,768	3.00 GiB	10.00 GiB	3.33×
65,536	6.00 GiB	20.00 GiB	3.33×
131,072	12.00 GiB	40.00 GiB	3.33×