Add model card

README.md

@@ -0,0 +1,50 @@
+---
+license: mit
+library_name: whisper.cpp
+pipeline_tag: automatic-speech-recognition
+tags:
+  - automatic-speech-recognition
+  - whisper
+  - ggml
+  - gguf
+  - mirror
+---
+
+# whisper.cpp GGML mirror
+
+An availability mirror of the GGML Whisper models used by the Friend macOS app.
+These are **byte-identical copies** of files published by
+[ggerganov/whisper.cpp](https://huggingface.co/ggerganov/whisper.cpp), pinned to a
+specific upstream commit and re-hosted so the app has a fallback if the primary
+becomes unavailable.
+
+Nothing here is retrained, requantized, or otherwise modified. If you are not the
+Friend app, **use the upstream repository instead** — it is the source of truth
+and is actively maintained.
+
+## Why this exists
+
+The app verifies every model it downloads against a pinned SHA-256 digest. A
+mirror lets a download that fails against the primary URL — transport error,
+deleted repo, or a digest mismatch — retry once from a second source before
+failing. Each file here is verified against the same pinned digest after upload.
+
+## Contents
+
+| File | Upstream |
+|---|---|
+| `ggml-tiny.bin` | ggerganov/whisper.cpp |
+| `ggml-base.bin` | ggerganov/whisper.cpp |
+| `ggml-small.bin` | ggerganov/whisper.cpp |
+| `ggml-medium-q5_0.bin` | ggerganov/whisper.cpp |
+| `ggml-large-v3.bin` | ggerganov/whisper.cpp |
+| `ggml-large-v3-turbo.bin` | ggerganov/whisper.cpp |
+
+## Attribution and licence
+
+- Original model: [Whisper](https://github.com/openai/whisper) by OpenAI, MIT licensed.
+- GGML conversion: [whisper.cpp](https://github.com/ggerganov/whisper.cpp) by
+  Georgi Gerganov and contributors, MIT licensed.
+
+This mirror is redistributed under the same MIT licence. All credit belongs to the
+original authors.