Is Qwen3-Coder-30B-A3B-Instruct or Mistral-MOE-4X7B-Dark-MultiVerse-Uncensored-Enhanced32-24B smaller?

Mistral-MOE-4X7B-Dark-MultiVerse-Uncensored-Enhanced32-24B is smaller at Q4_K_M: 13.61 GiB against 17.28 GiB.

Which uses less memory at long context?

Qwen3-Coder-30B-A3B-Instruct — its KV cache at 32K context is 1.3x smaller, because of how its attention layers cache.

Model comparison

Qwen3-Coder-30B-A3B-Instruct vs Mistral-MOE-4X7B-Dark-MultiVerse-Uncensored-Enhanced32-24B

At Q4_K_M, Mistral-MOE-4X7B-Dark-MultiVerse-Uncensored-Enhanced32-24B is the smaller download — 14,609,885,184 bytes against 18,556,689,568. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct's KV cache at 32K is 1.3× 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	Mistral-MOE-4X7B-Dark-MultiVerse-Uncensored-Enhanced32-24B
Parameters	30.5B	24.2B
Architecture	qwen3moe	llama
Layers	48	32
Native context	262,144	32,768
Mixture of experts	yes, 128 experts	yes, 4 experts
Quantizations published	46	33
Smallest quantization	7.46 GiB	8.24 GiB
Q4_K_M	17.28 GiB	13.61 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Qwen3-Coder-30B-A3B-Instruct	Mistral-MOE-4X7B-Dark-MultiVerse-Uncensored-Enhanced32-24B	Ratio
4,096	0.38 GiB	0.50 GiB	1.33×
8,192	0.75 GiB	1.00 GiB	1.33×
16,384	1.50 GiB	2.00 GiB	1.33×
32,768	3.00 GiB	4.00 GiB	1.33×
65,536	6.00 GiB	8.00 GiB	1.33×
131,072	12.00 GiB	16.00 GiB	1.33×