Documentation Index
Fetch the complete documentation index at: https://hyperwhisper.com/docs/llms.txt
Use this file to discover all available pages before exploring further.
HyperWhisper runs natively on macOS and Windows. Pick your platform below for minimum versions, architecture support, and hardware notes that affect local model performance.
Minimum requirements
- macOS 14 (Sonoma) or later
- Apple Silicon (M-series) or Intel — universal binary
- ~200 MB free disk for the app, plus space for any local models you download (78 MB to 3.1 GB each)
Apple Silicon vs Intel
| Feature | Apple Silicon (M1+) | Intel |
|---|
| App | Native | Native (universal binary) |
| Local Whisper / Parakeet | GPU + Neural Engine accelerated | CPU only |
| HyperWhisper Cloud | Same | Same |
Network
- HyperWhisper Cloud and BYOK providers require an internet connection.
- Local models transcribe entirely offline once downloaded.
- License activation requires a one-time network call.
See Network Filtering if you’re behind a corporate firewall.Minimum requirements
- Windows 10 or later (Windows 11 recommended)
- .NET 10 Runtime (installer prompts for it if missing)
- x64 or ARM64 (Snapdragon) — separate installers
- Visual C++ Redistributable 2015–2022 (auto-installed)
- DirectX 11 compatible GPU recommended for local model acceleration (any modern NVIDIA, AMD, or Intel GPU)
- ~300 MB free disk for the app, plus space for any local models (78 MB to 3.1 GB each)
x64 vs ARM64
| Feature | x64 | ARM64 |
|---|
| App | Native | Native |
| Whisper local models | Yes (DirectCompute GPU) | Not supported yet |
| Parakeet local models | Yes (DirectML GPU) | Yes |
| HyperWhisper Cloud | Yes | Yes |
GPU acceleration
Local engines use your GPU automatically when available and fall back to CPU if you don’t have a dedicated GPU or there’s not enough VRAM. See Local Models for per-model VRAM recommendations.