VibeVoice models
4 VibeVoice models on Hugging Face, from 323M to 8.7B parameters. At the precision each one is published in, the smallest needs about 1.4 GB of VRAM (VibeVoice-ASR-BitNet, cheapest live fit: V100) and the largest about 19.4 GB (VibeVoice-ASR, cheapest live fit: RTX A5000). The cheapest way to run VibeVoice-ASR-BitNet is $0.088/hr.
VibeVoice models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| VibeVoice-ASR-BitNet | 323M | F32 | 1.4 GB | V100 | $0.088/hr |
| VibeVoice-1.5B | 2.7B | BF16 | 6.0 GB | RTX 3060 | $0.110/hr |
| VibeVoice-ASR-HF | 8.3B | BF16 | 18.6 GB | RTX A5000 | $0.176/hr |
| VibeVoice-ASR | 8.7B | BF16 | 19.4 GB | RTX A5000 | $0.176/hr |
VRAM is for the precision the model is published in, with the same overhead and an 8,192-token context assumed on every page; see the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
More from Microsoft
- Florence-2 (5 models)
- Phi-4 (7 models)
- Phi-3.5 (4 models)
- Phi-3 (6 models)
- microsoft (6 models)
