Loading…
Loading…
Models / whisper-large
3 models (~1.54B to ~1.54B) · ↓ 10.1M downloads a month · from $0.06/h on GTX 1650 right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalopenai/whisper-large-v3 | vLLM | 9.3 GB | 13.1 GB | $0.13/h RTX A4000 |
| turboopenai/whisper-large-v3-turbo | vLLM | 1.6 GB | 3.9 GB | $0.09/h RTX 3060 |
| FP8 dynamic (RedHatAI)RedHatAI/whisper-large-v3-FP8-dynamic | vLLM | 1.7 GB | 4 GB | $0.09/h RTX 3060 |
| kb whisper large (KBLab)KBLab/kb-whisper-large | vLLM | 3.2 GB | 5.9 GB | $0.09/h RTX 3060 |
| turbo (onnx-community)onnx-community/whisper-large-v3-turbo | ONNX Runtime | 190 MB | 1.3 GB | $0.06/h GTX 1650 |
| turbo timestamped (onnx-community)onnx-community/whisper-large-v3-turbo_timestamped | ONNX Runtime | 190 MB | 1.3 GB | $0.06/h GTX 1650 |
| turbo gguf (xkeyC)xkeyC/whisper-large-v3-turbo-gguf | ONNX Runtime | 1.8 GB | 3.6 GB | $0.06/h GTX 1650 |
| turbo (TheStageAI)TheStageAI/thewhisper-large-v3-turbo | vLLM | 1.6 GB | 3.9 GB | $0.09/h RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalopenai/whisper-large-v2 | vLLM | 3.1 GB | 5.7 GB | $0.09/h RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalopenai/whisper-large | vLLM | 3.1 GB | 5.7 GB | $0.09/h RTX 3060 |
| Main (NbAiLab)NbAiLab/nb-whisper-large | vLLM | 3.1 GB | 5.7 GB | $0.09/h RTX 3060 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").