---
title: Falcon-OCR-bf16
canonical_url: "https://www.modelscope.cn/models/mlx-community/Falcon-OCR-bf16"
md_url: "https://www.modelscope.cn/models/mlx-community/Falcon-OCR-bf16.md"
repository: mlx-community/Falcon-OCR-bf16
last_updated: 2026-04-12
license: apache-2.0
pipeline_tag: image-text-to-text
tasks:
  - image-text-to-text
model_type:
  - falcon_ocr
architectures:
  - FalconOCRForCausalLM
base_model:
  - tiiuae/Falcon-OCR
base_model_relation: finetune
parameters: 270.2M
tensor_type:
  - BF16
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 17
stars: 0
tags:
  - mlx
  - ocr
  - vision-language
  - falcon
  - apple-silicon
---

# Falcon-OCR-bf16

> Falcon-OCR-bf16 - mlx-community 在 ModelScope 开源的模型。Falcon-OCR (MLX, bf16)

mlx-community/Falcon-OCR-bf16 是 ModelScope 魔搭社区上的 270.2M 参数image-text-to-text模型，采用 apache-2.0 许可，基于 tiiuae/Falcon-OCR 构建。

- **Repository**: mlx-community/Falcon-OCR-bf16
- **License**: apache-2.0
- **Tasks**: image-text-to-text
- **Parameters**: 270.2M
- **Base model**: tiiuae/Falcon-OCR
- **Tags**: mlx, ocr, vision-language, falcon, apple-silicon
- **Downloads**: 17
- **Stars**: 0
- **Last updated**: 2026-04-12

Source: https://www.modelscope.cn/models/mlx-community/Falcon-OCR-bf16

---

# Falcon-OCR (MLX, bf16)

MLX-converted weights of
[`tiiuae/Falcon-OCR`](https://huggingface.co/tiiuae/Falcon-OCR) for inference
on Apple Silicon via [`mlx-vlm`](https://github.com/Blaizzy/mlx-vlm).

## Source

- Base model: [tiiuae/Falcon-OCR](https://huggingface.co/tiiuae/Falcon-OCR)
  — Falcon Perception Team, Technology Innovation Institute (TII).
- License: Apache 2.0 (matches the upstream base model).
- Architecture: early-fusion vision-language model, 300M parameters.

## Conversion details

- Tool: `mlx_vlm.convert` (mlx-vlm 0.4.4).
- Dtype: `bfloat16`.
- Source revision: `3a4d95a8b0008f7430df30a82cf35e6c3b6bcb66`.
- `trust_remote_code=True` — the repository ships a custom
  `FalconOCRProcessor` / `FalconOCRForCausalLM` that is loaded via
  dynamic module import.

## Known caveat

`mlx_vlm.convert` raises
`AttributeError: 'FalconOCRProcessor' object has no attribute 'save_pretrained'`
at the very end of the conversion step. The weights and tokenizer are written
successfully before the error — so the artifacts uploaded here are complete
and `mlx_vlm.load(...)` / docling's `MlxVlmEngine` can consume them.

Tracked upstream: <https://github.com/Blaizzy/mlx-vlm/issues>.

## Usage

```python
from mlx_vlm import load, generate
model, processor = load("mlx-community/Falcon-OCR-bf16", trust_remote_code=True)
output = generate(model, processor, prompt="", image=["path/to/page.png"])
print(output)
```

## Attribution

All credit for the underlying model goes to the Falcon Perception Team at TII.
Cite the [model card](https://huggingface.co/tiiuae/Falcon-OCR) for academic
references.