distil-whisper models
2 distil-whisper models on Hugging Face, from 166M to 756M parameters. At the precision each one is published in, the smallest needs about 0.4 GB of VRAM (distil-small.en, cheapest live fit: V100) and the largest about 1.7 GB (distil-large-v3, cheapest live fit: V100). The cheapest way to run distil-small.en is $0.088/hr.
distil-whisper models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| distil-small.en | 166M | F16 | 0.4 GB | V100 | $0.088/hr |
| distil-large-v3 | 756M | F16 | 1.7 GB | V100 | $0.088/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.
