Model comparison
gemma-4-31B-it-DFlash vs gemma-3-1b-it
At Q4_K_M, gemma-3-1b-it is the smaller download — 806,058,240 bytes against 911,587,296.
From the file· summed bytes, KV per layer
Side by side
| gemma-4-31B-it-DFlash | gemma-3-1b-it | |
|---|---|---|
| Parameters | 1.5B | 1000M |
| Architecture | dflash | gemma3 |
| Layers | 5 | 26 |
| Native context | 262,144 | 32,768 |
| Mixture of experts | no | no |
| Quantizations published | 12 | 28 |
| Smallest quantization | 0.54 GiB | 0.52 GiB |
| Q4_K_M | 0.85 GiB | 0.75 GiB |
| Licence | apache-2.0 | gemma |
KV cache by context
the term that decides long-context viability
| Context | gemma-4-31B-it-DFlash | gemma-3-1b-it | Ratio |
|---|---|---|---|
| 4,096 | 0.05 GiB | 0.04 GiB | 1.47× |
| 8,192 | 0.07 GiB | 0.05 GiB | 1.33× |
| 16,384 | 0.10 GiB | 0.08 GiB | 1.21× |
| 32,768 | 0.16 GiB | 0.15 GiB | 1.12× |
| 65,536 | 0.29 GiB | 0.27 GiB | 1.06× |
| 131,072 | 0.54 GiB | 0.52 GiB | 1.03× |