Model comparison
VibeVoice-1.5B vs s2-pro
At Q4_K_M, VibeVoice-1.5B is the smaller download — 3,536,105,184 bytes against 3,566,165,088.
From the file· summed bytes, KV per layer
Side by side
| VibeVoice-1.5B | s2-pro | |
|---|---|---|
| Parameters | 2.7B | 4.6B |
| Architecture | vibevoice-asr | fish-speech |
| Layers | — | 36 |
| Native context | — | — |
| Mixture of experts | no | yes, 1 experts |
| Quantizations published | 20 | 7 |
| Smallest quantization | 1.76 GiB | 2.40 GiB |
| Q4_K_M | 3.29 GiB | 3.32 GiB |
| Licence | mit | other |
KV cache by context
the term that decides long-context viability
| Context | VibeVoice-1.5B | s2-pro | Ratio |
|---|---|---|---|
| 4,096 | — | — | — |
| 8,192 | — | — | — |
| 16,384 | — | — | — |
| 32,768 | — | — | — |
| 65,536 | — | — | — |
| 131,072 | — | — | — |