FFmpeg
Video and audio conversion, trimming, merging, subtitles, filters, frame interpolation, GIFs and metadata removal.
≈ 105 MB download
yt-dlp + Deno
YouTube metadata and media retrieval for focused tools and batch workspaces.
≈ 58 MB download
whisper.cpp
Local speech recognition. Tiny, Base, Small and Large v3 Turbo quantized models fetched individually.
31 MB – 548 MB per model
Real-ESRGAN ncnn
Vulkan image and video-frame enhancement to 2K or 4K on a capable graphics adapter.
≈ 43 MB download
ONNX Runtime + U2NetP
CPU inference for the lightweight background model, downloaded once and reused.
≈ 4.4 MB model
Windows OCR
Built into the operating system. Text recognition for scans, images and PDFs with nothing extra to fetch.
Ships with Windows