Is Llama-3.2-8X3B-MOE-Dark-Champion-Instruct-uncensored-abliterated-18.4B or Qwen3-Coder-30B-A3B-Instruct smaller?

Llama-3.2-8X3B-MOE-Dark-Champion-Instruct-uncensored-abliterated-18.4B is smaller at Q4_K_M: 10.54 GiB against 17.28 GiB.

Which uses less memory at long context?

Qwen3-Coder-30B-A3B-Instruct — its KV cache at 32K context is 1.2x smaller, because of how its attention layers cache.

Model comparison

Llama-3.2-8X3B-MOE-Dark-Champion-Instruct-uncensored-abliterated-18.4B vs Qwen3-Coder-30B-A3B-Instruct

At Q4_K_M, Llama-3.2-8X3B-MOE-Dark-Champion-Instruct-uncensored-abliterated-18.4B is the smaller download — 11,312,942,112 bytes against 18,556,689,568. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct's KV cache at 32K is 1.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Llama-3.2-8X3B-MOE-Dark-Champion-Instruct-uncensored-abliterated-18.4B	Qwen3-Coder-30B-A3B-Instruct
Parameters	18.4B	30.5B
Architecture	llama	qwen3moe
Layers	28	48
Native context	131,072	262,144
Mixture of experts	yes, 8 experts	yes, 128 experts
Quantizations published	31	46
Smallest quantization	6.56 GiB	7.46 GiB
Q4_K_M	10.54 GiB	17.28 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Llama-3.2-8X3B-MOE-Dark-Champion-Instruct-uncensored-abliterated-18.4B	Qwen3-Coder-30B-A3B-Instruct	Ratio
4,096	0.44 GiB	0.38 GiB	1.17×
8,192	0.88 GiB	0.75 GiB	1.17×
16,384	1.75 GiB	1.50 GiB	1.17×
32,768	3.50 GiB	3.00 GiB	1.17×
65,536	7.00 GiB	6.00 GiB	1.17×
131,072	14.00 GiB	12.00 GiB	1.17×