---
language:
  - zh
  - en
  - ja
  - ko
  - fr
  - de
  - es
  - ru
  - ar
  - hi
  - it
  - pt
  - tr
  - nl
library_name: mlx
base_model: Qwen/Qwen3-ASR-1.7B
tags:
  - mlx
  - qwen3
  - asr
  - speech-recognition
  - quantized
  - 8bit
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---

# Qwen3-ASR-1.7B-Q8

8-bit quantized version of [Qwen/Qwen3-ASR-1.7B](https://huggingface.co/Qwen/Qwen3-ASR-1.7B) for MLX.

## Quantization

- Bits: 8
- Group size: 64
- Format: MLX safetensors
- Quantized using [mlx-qwen3-asr](https://github.com/moona3k/mlx-qwen3-asr)

## Usage

```python
from mlx_qwen3_asr import Session

session = Session("illitan/Qwen3-ASR-1.7B-Q8")
result = session.transcribe("audio.wav")
print(result.text)
```

## Source

Quantized from [Qwen/Qwen3-ASR-1.7B](https://huggingface.co/Qwen/Qwen3-ASR-1.7B) using `nn.quantize(model, bits=8, group_size=64)`.