Is v6-Finch-14B-HF or Qwen3-30B-A3B-Thinking-2507 smaller?

v6-Finch-14B-HF is smaller at Q4_K_M: 8.17 GiB against 17.28 GiB.

Which uses less memory at long context?

Qwen3-30B-A3B-Thinking-2507 — its KV cache at 32K context is 10.2x smaller, because of how its attention layers cache.

Model comparison

v6-Finch-14B-HF vs Qwen3-30B-A3B-Thinking-2507

At Q4_K_M, v6-Finch-14B-HF is the smaller download — 8,767,898,176 bytes against 18,556,685,824. At long context the gap widens: Qwen3-30B-A3B-Thinking-2507's KV cache at 32K is 10.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	v6-Finch-14B-HF	Qwen3-30B-A3B-Thinking-2507
Parameters	14.1B	30.5B
Architecture	rwkv6	qwen3moe
Layers	61	48
Native context	—	262,144
Mixture of experts	no	yes, 128 experts
Quantizations published	20	51
Smallest quantization	5.10 GiB	7.05 GiB
Q4_K_M	8.17 GiB	17.28 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	v6-Finch-14B-HF	Qwen3-30B-A3B-Thinking-2507	Ratio
4,096	3.81 GiB	0.38 GiB	10.17×
8,192	7.63 GiB	0.75 GiB	10.17×
16,384	15.25 GiB	1.50 GiB	10.17×
32,768	30.50 GiB	3.00 GiB	10.17×
65,536	61.00 GiB	6.00 GiB	10.17×
131,072	122.00 GiB	12.00 GiB	10.17×