---
library_name: mlx
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen3.5-35B-A3B/blob/main/LICENSE
pipeline_tag: image-text-to-text
base_model: Qwen/Qwen3.5-35B-A3B
tags:
- mlx
---

[Qwen3.5-35B-A3B](https://huggingface.co/Qwen/Qwen3.5-35B-A3B) optimized for MLX.
This quant supports image input and requires a vision-enabled MLX server.

For the non-vision model: https://huggingface.co/spicyneuron/Qwen3.5-35B-A3B-MLX-4.8bit

**EDIT:** Updated chat template to enable better prompt caching.

# Usage

```sh
# Start server at http://localhost:8080/chat/completions
uvx --from mlx-vlm --with torchvision \
  mlx_vlm.server \
  --host 127.0.0.1 \
  --port 8080 \
  --model spicyneuron/Qwen3.5-35B-A3B-MLX-4.9bit-vision
```

# Methodology

Quantized using a custom script inspired by Unsloth/AesSedai/ubergarm style mixed-precision GGUFs.
MLX quantization options differ than llama.cpp, but the principles are the same:

- Sensitive layers like MoE routing, attention, and output embeddings get higher precision
- More tolerant layers like MoE experts get lower precision