---
library_name: mlx
tags:
- mlx
- oq
- quantized
base_model:
- deepreinforce-ai/Ornith-1.0-35B
---

# Ornith-1.0-35B-MLX-oQ3.5-fp16

I told Claude Opus to analyse Unsloth-Qwen3.6-35B-A3B-UD-3bit and Ornith-1.0-35B-MLX-oQ4 layers then quantized from the original model.

## Quantization details

- **Model type**: qwen3_5_moe
- **Bits**: 3
- **Group size**: 64
- **Format**: MLX safetensors

## Benchmark

Tested via omlx
- **Humaneval**: 97/100
- **MBPP**: 92/100
- **Livecodebench**: 45/100