Model comparison
Delphi-25B-SimpleRL-Math 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 13.6× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| Delphi-25B-SimpleRL-Math | gemma-4-12B-it-qat-q4_0-unquantized | |
|---|---|---|
| Parameters | 25.0B | 12.0B |
| Architecture | qwen3 | gemma4 |
| Layers | 51 | 48 |
| Native context | 16,384 | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 34 | 2 |
| Smallest quantization | 5.31 GiB | 6.50 GiB |
| Q4_K_M | 14.12 GiB | — |
| Licence | — | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Delphi-25B-SimpleRL-Math | gemma-4-12B-it-qat-q4_0-unquantized | Ratio |
|---|---|---|---|
| 4,096 | 4.18 GiB | 0.72 GiB | 5.82× |
| 8,192 | 8.37 GiB | 0.97 GiB | 8.64× |
| 16,384 | 16.73 GiB | 1.47 GiB | 11.39× |
| 32,768 | 33.47 GiB | 2.47 GiB | 13.56× |
| 65,536 | 66.94 GiB | 4.47 GiB | 14.98× |
| 131,072 | 133.88 GiB | 8.47 GiB | 15.81× |