Model comparison
dolphin-2.6-mixtral-8x7b 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 1.6× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| dolphin-2.6-mixtral-8x7b | gemma-4-12B-it-qat-q4_0-unquantized | |
|---|---|---|
| Parameters | 46.7B | 12.0B |
| Architecture | llama | gemma4 |
| Layers | 32 | 48 |
| Native context | 32,768 | 262,144 |
| Mixture of experts | yes, 8 experts | no |
| Quantizations published | 43 | 2 |
| Smallest quantization | 9.15 GiB | 6.50 GiB |
| Q4_K_M | 24.63 GiB | — |
| Licence | apache-2.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | dolphin-2.6-mixtral-8x7b | gemma-4-12B-it-qat-q4_0-unquantized | Ratio |
|---|---|---|---|
| 4,096 | 0.50 GiB | 0.72 GiB | 1.44× |
| 8,192 | 1.00 GiB | 0.97 GiB | 1.03× |
| 16,384 | 2.00 GiB | 1.47 GiB | 1.36× |
| 32,768 | 4.00 GiB | 2.47 GiB | 1.62× |
| 65,536 | 8.00 GiB | 4.47 GiB | 1.79× |
| 131,072 | 16.00 GiB | 8.47 GiB | 1.89× |