v1.0.0

whisper-large-v3-ggml v1.0.0

What's included

• ggml-large-v3-q8_0.bin: Q8 quantized build of OpenAI's Whisper Large v3 model.
• checksums.txt: SHA-256 digest for verifying downloads.

Provenance

Base weights downloaded from https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3.bin and quantized with whisper.cpp's quantize q8_0 utility.

Verification

curl -L -o ggml-large-v3-q8_0.bin \
  https://github.com/sergheinenov/whisper-large-v3-ggml/releases/download/v1.0.0/ggml-large-v3-q8_0.bin
curl -L -o checksums.txt \
  https://github.com/sergheinenov/whisper-large-v3-ggml/releases/download/v1.0.0/checksums.txt
shasum -a 256 -c checksums.txt

SHA-256:

24bc434f372355688ab9a623077a63e5361a1c41f4d8d648977e39f9b060f09e

Notes

• Keep the binary out of git history; attach it only as a release asset.
• Recompute the checksum if you regenerate the quantized file.