---
title: Ornith-1.0-35B-3bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/Ornith-1.0-35B-3bit"
md_url: "https://www.modelscope.cn/models/mlx-community/Ornith-1.0-35B-3bit.md"
repository: mlx-community/Ornith-1.0-35B-3bit
last_updated: 2026-06-28
license: mit
pipeline_tag: image-text-to-text
tasks:
  - image-text-to-text
model_type:
  - qwen3_5_moe
architectures:
  - Qwen3_5MoeForConditionalGeneration
base_model:
  - deepreinforce-ai/Ornith-1.0-35B
base_model_relation: quantized
parameters: 4.8B
tensor_type:
  - U32
  - BF16
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 86
stars: 0
tags:
  - mlx
  - mlx-vlm
  - vlm
  - moe
  - quantized
---

# Ornith-1.0-35B-3bit

> Ornith-1.0-35B-3bit - mlx-community 在 ModelScope 开源的模型。3-bit (group size 64, 3.662 bits/weight) MLX quantization of deepreinforce-ai/Ornith-1.0-35B, produced with mlx-vlm 0.6.3. Full multimodal: the vision encoder is preserved and quantized alongside the…

mlx-community/Ornith-1.0-35B-3bit 是 ModelScope 魔搭社区上的 4.8B 参数image-text-to-text模型，采用 mit 许可，基于 deepreinforce-ai/Ornith-1.0-35B 构建。

- **Repository**: mlx-community/Ornith-1.0-35B-3bit
- **License**: mit
- **Tasks**: image-text-to-text
- **Parameters**: 4.8B
- **Base model**: deepreinforce-ai/Ornith-1.0-35B
- **Tags**: mlx, mlx-vlm, vlm, moe, quantized
- **Downloads**: 86
- **Stars**: 0
- **Last updated**: 2026-06-28

Source: https://www.modelscope.cn/models/mlx-community/Ornith-1.0-35B-3bit

---

# Ornith-1.0-35B-3bit

3-bit (group size 64, **3.662 bits/weight**) MLX quantization of
[deepreinforce-ai/Ornith-1.0-35B](https://huggingface.co/deepreinforce-ai/Ornith-1.0-35B),
produced with `mlx-vlm` 0.6.3. Full multimodal: the vision encoder is preserved and
quantized alongside the language model. For Apple Silicon. Runs in `mlx-vlm` or any MLX app.

This is the smallest variant (≈16 GB). It stays coherent on both vision and reasoning, but is the
most aggressive precision — expect some quality loss versus 4-bit and up.

## Conversion note (MoE expert fusion)

Ornith stores its 256 MoE experts unfused (per-expert), but mlx-vlm's `qwen3_5_moe` loader expects
them fused/batched. A `sanitize` monkeypatch was required to stack the experts before conversion; without it the
conversion failed. This is a standard mlx-vlm 3-bit quant.

## Usage

```bash
uvx --from mlx-vlm mlx_vlm.generate \
  --model mlx-community/Ornith-1.0-35B-3bit --image image.png \
  --prompt "Describe this image." --max-tokens 512
```

```python
from mlx_vlm import load, generate
model, processor = load("mlx-community/Ornith-1.0-35B-3bit")
```

## Conversion check

Smoke-tested after conversion: coherent on both an image prompt (correctly read an evaluation bar
chart) and a text reasoning prompt (`17 * 24` solved as 408 with correct step-by-step work), no
repetition loop. **125.3 tok/s** generation, **946.2 tok/s** prompt, **peak 18.1 GB** on a
Macbook Pro M5 Max 128GB 40 GPU.

Refer to the original model card for architecture, benchmarks, license, and intended use.