faster-whisper
SYSTRANA reimplementation of OpenAI's Whisper on the CTranslate2 inference engine, up to four times faster than the reference code at the same accuracy and using less memory, with 8-bit quantization on CPU and GPU, and batched inference.
Openness
5 medium confidence- license
- MIT(OSI)
- source
- public
- core features withheld
- no
faster-whisper is MIT-licensed and the published repository is the whole library; nothing is sold alongside it or withheld. SYSTRAN publishes no statement about commercial offerings, so that reading comes from the repository alone.
- https://raw.githubusercontent.com/SYSTRAN/faster-whisper/HEAD/LICENSE recorded 2026-09-27
LICENSE body: "MIT License" / "Copyright (c) 2023 SYSTRAN".
- https://raw.githubusercontent.com/SYSTRAN/faster-whisper/HEAD/README.md recorded 2026-09-27
README describes a pip-installable library and its benchmarks; it names no paid tier, license key or enterprise edition.
Adoption
4 high confidencePyPI downloads of faster-whisper. Installs pulled in as a dependency of other tools count the same as direct ones.
- https://pypistats.org/api/packages/faster-whisper/recent recorded 2026-09-27
"last_month":8271086 for faster-whisper.
Capability
3 medium confidenceA fast runtime, but for one model family and one task, and it trains nothing.
- https://raw.githubusercontent.com/SYSTRAN/faster-whisper/HEAD/README.md recorded 2026-09-27
"This implementation is up to 4 times faster than [openai/whisper](https://github.com/openai/whisper) for the same accuracy while using less memory."
Verified 2026-09-27