Omi Med STT v1 MLX
Full-precision Apple Silicon / MLX export of Omi Med STT v1.
For most Mac users, the smaller q8 export is the recommended default. Use this repo when you specifically want the full MLX weights.
Quickstart
pip install -U "omi-med-stt[mlx]"
omi-med-stt audio.wav --runtime mlx --model omi-health/omi-med-stt-v1-mlx
Evaluation
The public weights are unchanged. The full-precision export remains available for experimentation, but it is not the selected or separately re-qualified September 2026 Apple runtime. The supported Mac path is the q8 artifact.
For context, these are the currently qualified runtime results on the same frozen 1,513-clip, 7.18-hour medical benchmark and scorer:
| Runtime artifact | Platform | WER | M-WER | Drug M-WER | Medical recall |
|---|---|---|---|---|---|
| NeMo canonical | NVIDIA CUDA (L4, BF16) | 6.54% | 2.23% | 4.75% | 97.77% |
| MLX q8 (selected) | Apple Silicon (M4 Max) | 6.65% | 2.12% | 4.52% | 97.88% |
| GGUF q8_0 | Linux/Windows CPU | 7.10% | 2.16% | 4.30% | 97.84% |
Do not attribute the q8 result to this full-precision artifact. Choose this
repository only when you specifically need the unquantized MLX weights; use
omi-med-stt-v1-mlx-q8
for the qualified Mac runtime.
Compatibility
This is not a drop-in parakeet-mlx checkpoint. Omi Med STT v1 includes a
medical adapter, and the supported Mac path is the omi-med-stt CLI.
Links
- Canonical model:
omi-health/omi-med-stt-v1 - Mac q8 default:
omi-health/omi-med-stt-v1-mlx-q8 - CPU GGUF export:
omi-health/omi-med-stt-v1-gguf - Runtime CLI:
Omi-Health/omi-med-stt-runtime - Broader evaluation and product context: omi.health/research/omi-med-stt
Safety
Omi Med STT v1 is speech-to-text only. It is not a diagnostic, triage, prescribing, or clinical decision model, and it is not clinically validated. Transcripts must be reviewed before any clinical use.