Model comparison
14B vs gemma-4-12B-it-qat-q4_0-unquantized
These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-12B-it-qat-q4_0-unquantized's KV cache at 32K is 10.1× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| 14B | gemma-4-12B-it-qat-q4_0-unquantized | |
|---|---|---|
| Parameters | 14.2B | 12.0B |
| Architecture | llama | gemma4 |
| Layers | 40 | 48 |
| Native context | 8,192 | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 5 | 2 |
| Smallest quantization | 7.62 GiB | 6.50 GiB |
| Q4_K_M | — | — |
| Licence | wtfpl | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | 14B | gemma-4-12B-it-qat-q4_0-unquantized | Ratio |
|---|---|---|---|
| 4,096 | 3.13 GiB | 0.72 GiB | 4.35× |
| 8,192 | 6.25 GiB | 0.97 GiB | 6.45× |
| 16,384 | 12.50 GiB | 1.47 GiB | 8.51× |
| 32,768 | 25.00 GiB | 2.47 GiB | 10.13× |
| 65,536 | 50.00 GiB | 4.47 GiB | 11.19× |
| 131,072 | 100.00 GiB | 8.47 GiB | 11.81× |