Parakeet models
7 Parakeet models on Hugging Face, from 609M to 1.1B parameters. At the precision each one is published in, the smallest needs about 1.4 GB of VRAM (parakeet-ultra, cheapest live fit: V100) and the largest about 4.8 GB (parakeet-ctc-1.1b, cheapest live fit: V100). The cheapest way to run parakeet-ultra is $0.088/hr.
Parakeet models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| parakeet-ctc-0.6b | 609M | F32 | 2.7 GB | V100 | $0.088/hr |
| parakeet-rnnt-0.6b | 617M | F32 | 2.8 GB | V100 | $0.088/hr |
| parakeet-tdt-0.6b-v2 | 618M | F32 | 2.8 GB | V100 | $0.088/hr |
| parakeet-tdt-0.6b-v3 | 627M | F32 | 2.8 GB | V100 | $0.088/hr |
| parakeet-tdt-0.6b-v3 | 627M | F32 | 2.8 GB | V100 | $0.088/hr |
| parakeet-ultra | 627M | F16 | 1.4 GB | V100 | $0.088/hr |
| parakeet-ctc-1.1b | 1.1B | F32 | 4.8 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.
More from NVIDIA
- Nemotron 3 (16 models)
- Cosmos (6 models)
- Nemotron Nano (4 models)
- Nemotron (5 models)
- Nemotron Labs (4 models)
- Nemotron-H (3 models)
