Model comparison
Apriel-1.6-15b-Thinker vs gemma-4-31B-it
At Q4_K_M, Apriel-1.6-15b-Thinker is the smaller download — 8,785,478,144 bytes against 18,323,733,440.
From the file· summed bytes, KV per layer
Side by side
| Apriel-1.6-15b-Thinker | gemma-4-31B-it | |
|---|---|---|
| Parameters | 14.9B | 31.3B |
| Architecture | llama | gemma4 |
| Layers | 48 | 60 |
| Native context | 262,400 | 262,144 |
| Mixture of experts | no | no |
| Quantizations published | 60 | 55 |
| Smallest quantization | 3.22 GiB | 7.95 GiB |
| Q4_K_M | 8.18 GiB | 17.07 GiB |
| Licence | mit | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Apriel-1.6-15b-Thinker | gemma-4-31B-it | Ratio |
|---|---|---|---|
| 4,096 | 0.75 GiB | 1.80 GiB | 2.40× |
| 8,192 | 1.50 GiB | 2.42 GiB | 1.61× |
| 16,384 | 3.00 GiB | 3.67 GiB | 1.22× |
| 32,768 | 6.00 GiB | 6.17 GiB | 1.03× |
| 65,536 | 12.00 GiB | 11.17 GiB | 1.07× |
| 131,072 | 24.00 GiB | 21.17 GiB | 1.13× |