Model comparison
Nethena-20B 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 15.7× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| Nethena-20B | gemma-4-12B-it-qat-q4_0-unquantized | |
|---|---|---|
| Parameters | 20.0B | 12.0B |
| Architecture | llama | gemma4 |
| Layers | 62 | 48 |
| Native context | 4,096 | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 12 | 2 |
| Smallest quantization | 7.74 GiB | 6.50 GiB |
| Q4_K_M | 11.22 GiB | — |
| Licence | cc-by-nc-4.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Nethena-20B | gemma-4-12B-it-qat-q4_0-unquantized | Ratio |
|---|---|---|---|
| 4,096 | 4.84 GiB | 0.72 GiB | 6.74× |
| 8,192 | 9.69 GiB | 0.97 GiB | 10.00× |
| 16,384 | 19.38 GiB | 1.47 GiB | 13.19× |
| 32,768 | 38.75 GiB | 2.47 GiB | 15.70× |
| 65,536 | 77.50 GiB | 4.47 GiB | 17.34× |
| 131,072 | 155.00 GiB | 8.47 GiB | 18.30× |