Is gemma-4-12B-coder-fable5-composer2.5-v1-bf16 or Qwen3-8B smaller?

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

Which uses less memory at long context?

gemma-4-12B-coder-fable5-composer2.5-v1-bf16 — its KV cache at 32K context is 1.8x smaller, because of how its attention layers cache.

Model comparison

gemma-4-12B-coder-fable5-composer2.5-v1-bf16 vs Qwen3-8B

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-12B-coder-fable5-composer2.5-v1-bf16's KV cache at 32K is 1.8× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gemma-4-12B-coder-fable5-composer2.5-v1-bf16	Qwen3-8B
Parameters	12.0B	8.2B
Architecture	gemma4	qwen3
Layers	48	36
Native context	262,144	40,960
Mixture of experts	no	no
Quantizations published	24	51
Smallest quantization	2.78 GiB	2.12 GiB
Q4_K_M	—	4.68 GiB
Licence	gemma	apache-2.0

KV cache by context

the term that decides long-context viability

Context	gemma-4-12B-coder-fable5-composer2.5-v1-bf16	Qwen3-8B	Ratio
4,096	0.72 GiB	0.56 GiB	1.28×
8,192	0.97 GiB	1.13 GiB	1.16×
16,384	1.47 GiB	2.25 GiB	1.53×
32,768	2.47 GiB	4.50 GiB	1.82×
65,536	4.47 GiB	9.00 GiB	2.01×
131,072	8.47 GiB	18.00 GiB	2.13×