Model comparison
gemma-4-12B-it-qat-q4_0-unquantized-assistant vs ced-base
These two publish different quantization sets; the table below has the exact sizes.
From the file· summed bytes, KV per layer
Side by side
| gemma-4-12B-it-qat-q4_0-unquantized-assistant | ced-base | |
|---|---|---|
| Parameters | 423M | 86M |
| Architecture | gemma4-assistant | ced |
| Layers | 4 | — |
| Native context | 262,144 | — |
| Mixture of experts | no | no |
| Quantizations published | 3 | 3 |
| Smallest quantization | 0.30 GiB | 0.12 GiB |
| Q4_K_M | — | — |
| Licence | apache-2.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | gemma-4-12B-it-qat-q4_0-unquantized-assistant | ced-base | Ratio |
|---|---|---|---|
| 4,096 | 0.07 GiB | — | — |
| 8,192 | 0.10 GiB | — | — |
| 16,384 | 0.16 GiB | — | — |
| 32,768 | 0.29 GiB | — | — |
| 65,536 | 0.54 GiB | — | — |
| 131,072 | 1.04 GiB | — | — |