---
language:
- en
license: apache-2.0
pipeline_tag: automatic-speech-recognition
library_name: mlx
tags:
- mlx
- automatic-speech-recognition
- speech-to-text
- audio
- cohere_asr
- quantized
- 8bit
base_model:
- CohereLabs/cohere-transcribe-03-2026
---

# cohere-transcribe-03-2026-mlx-8bit

Quantized MLX weights for **beshkenadze/cohere-transcribe-03-2026-mlx-fp16**.

## Variant

- Precision: **8-bit**
- Quantization mode: `affine`
- Group size: `64`

## Files

- `model.safetensors`
- `config.json`
- `tokenizer.model`
- `tokenizer_config.json`
- `preprocessor_config.json`
- `special_tokens_map.json`
- `key_map.json`
- `conversion_summary.json`

## Repo-sample benchmark

Sample: `Tests/media/conversational_a.wav`

- Generation TPS: **352.9**
- Peak memory: **2.87 GB**
- Output: `Coffee's story likely begins in Ethiopia, where legend tells of a goat herder named Kaldi, who noticed his goats became energetic after eating red berries from a particular bush; curious, he tried them himself and felt invigorated.`

## Parity note

This checkpoint has been re-validated against the current Swift and Python MLX runtimes.

Verified semantic parity on an English fixture:

> `This is a test recording in English. I am speaking clearly at a normal speed. Please transcribe this sentence exactly as I said.`

Matched across:

- Swift MLX fp16
- Swift MLX 8-bit
- Python MLX fp16
- Python MLX 8-bit
- official CUDA reference path (`transformers` native Cohere ASR)

## Quality note

Matches fp16 on the repo sample while reducing memory substantially.

## Notes

- Generated from the Swift-compatible fp16 checkpoint `beshkenadze/cohere-transcribe-03-2026-mlx-fp16`.
- This repository contains inference artifacts only. Refer to the upstream Cohere model card and license for original model details.