Back to Gap Map Model components / Speech & audio

whisper.cpp

open source / Overall score: 3.0

A dependency-free C/C++ port of Whisper on the ggml tensor library, with integer quantization and GPU backends including Metal, CUDA and Vulkan. It runs on desktops, servers, iOS, Android, WebAssembly and the Raspberry Pi.

There is no official Python package. The converted models on the Hub (ggerganov/whisper.cpp) report no download count, so adoption reads GitHub stars.

Openness

5 medium confidence

5.0

license: MIT(OSI)
source: public
core features withheld: no

whisper.cpp is MIT-licensed and the repository is the whole program. Its maintainer, ggml.ai, says the library and its related projects are freely available under MIT and that commercial extensions are at most a future possibility, so nothing is withheld today.

https://ggml.ai/ recorded 2026-09-27
Company page, under "Open Core": "The library and related projects are freely available under the MIT license." and "In the future we may choose to develop extensions that are licensed for commercial use".
https://raw.githubusercontent.com/ggml-org/whisper.cpp/HEAD/LICENSE recorded 2026-09-27
LICENSE body: "MIT License" / "Copyright (c) 2023-2026 The ggml authors".
https://raw.githubusercontent.com/ggml-org/whisper.cpp/HEAD/README.md recorded 2026-09-27
README: "Plain C/C++ implementation without dependencies"; builds from the repository for every listed platform, with no paid tier or license key mentioned.

Adoption

3 low confidence

3.0

GitHub stars, since whisper.cpp publishes no package and its Hub model repository reports zero downloads. Stars measure attention rather than use, and the stars scale stops at 3.

https://huggingface.co/api/models/ggerganov/whisper.cpp recorded 2026-09-27
"downloads":0 for ggerganov/whisper.cpp.
https://ungh.cc/repos/ggml-org/whisper.cpp recorded 2026-09-27
Repository record for ggml-org/whisper.cpp: "stars":53946.

Capability

3 medium confidence

3.0

The most portable Whisper runtime, but still one model family and one task.

https://raw.githubusercontent.com/ggml-org/whisper.cpp/HEAD/README.md recorded 2026-09-27
"Apple Silicon first-class citizen - optimized via ARM NEON, Accelerate framework, Metal and [Core ML](#core-ml-support)"; supported platforms include "[x] [WebAssembly](examples/whisper.wasm)".

Verified 2026-09-27