Fact · Speech recognition
The openai-whisper command line runs on an NVIDIA GPU (CUDA) only when PyTorch can detect one, and otherwise falls back to the CPU.
Checked against its source on October 17, 2026
In the source's own words
“parser.add_argument("--device", default="cuda" if torch.cuda.is_available() else "cpu", help="device to use for PyTorch inference")”
whisper/transcribe.py: decoding defaults and command-line options (openai/whisper), OpenAI, GitHub (openai/whisper), 2026, Command-line options (--device).
Open the original- What it measures
- The default is set once, when the program starts. If the GPU is not detected, the run silently uses the CPU, which is slower.
- What it does not show
- If a GPU is installed but not used, the usual cause is a PyTorch build without CUDA support. Installing a CUDA build of PyTorch changes the default.