Model comparison
GigaAM-v3 vs parakeet-tdt-0.6b-v2
At Q4_K_M, GigaAM-v3 is the smaller download — 182,497,888 bytes against 475,491,840.
From the file· summed bytes, KV per layer
Side by side
| GigaAM-v3 | parakeet-tdt-0.6b-v2 | |
|---|---|---|
| Parameters | 223M | 618M |
| Architecture | gigaam | parakeet |
| Layers | — | — |
| Native context | — | — |
| Mixture of experts | no | no |
| Quantizations published | 18 | 8 |
| Smallest quantization | 0.17 GiB | 0.37 GiB |
| Q4_K_M | 0.17 GiB | 0.44 GiB |
| Licence | mit | cc-by-4.0 |
KV cache by context
the term that decides long-context viability
| Context | GigaAM-v3 | parakeet-tdt-0.6b-v2 | Ratio |
|---|---|---|---|
| 4,096 | — | — | — |
| 8,192 | — | — | — |
| 16,384 | — | — | — |
| 32,768 | — | — | — |
| 65,536 | — | — | — |
| 131,072 | — | — | — |