4-bit Open on Hugging Face
Parameters
—
Weights (est.)
—
Mac memory (est.)
—
Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.
Run it
With mlx-vlm on an Apple silicon Mac:
pip install mlx-vlmpython -m mlx_vlm.generate --model mlx-community/olmOCR-2-7B-1025-4bit --image photo.jpg --prompt "Describe this image"
Details
- Base model
- Qwen/Qwen2.5-VL-7B-Instruct (quantized)
- Vendor
- Qwen
- Task
- Vision + text
- Quantisation
- 4-bit
- Licence
- apache-2.0
- Downloads
- 1.1K in the last 30 days · 2.4K all time · 0 likes
- Published
- 2025-10-23 · updated 2025-10-23 (1y ago)
- Languages
- en
- Tags
- qwen2_5_vl
Other MLX builds of Qwen/Qwen2.5-VL-7B-Instruct
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Qwen2.5-VL-7B-Instruct-3bit | 3-bit | ≈ 4.7 GB | 8 GB+ | 173 |
| Holo1-7B-3bit | 3-bit | ≈ 4.7 GB | 8 GB+ | 24 |
| Cosmos-Reason1-7B-4bit | 4-bit | ≈ 4.9 GB | 8 GB+ | 37 |
| Jedi-7B-1080p-4bit | 4-bit | ≈ 4.9 GB | 8 GB+ | 23 |
| VL-Rethinker-7B-4bit | 4-bit | ≈ 4.9 GB | 8 GB+ | 20 |
| Qwen2.5-VL-7B-Instruct-4bit | 4-bit | ≈ 5.6 GB | 16 GB+ | 21K |
| Qwen2.5-VL-7B-Instruct-6bit | 6-bit | ≈ 6.9 GB | 16 GB+ | 166 |
| Qwen2.5-VL-7B-Instruct-8bit | 8-bit | ≈ 8.9 GB | 16 GB+ | 2.5K |
| olmOCR-2-7B-1025-bf16 | bf16 | ≈ 16.6 GB | 24 GB+ | 1.0K |
| Qwen2.5-VL-7B-Instruct-bf16 | bf16 | ≈ 16.6 GB | 24 GB+ | 174 |
| holo1.5-7b-mlx | — | ≈ 16.6 GB | 24 GB+ | 36 |
| Holo1-7B-bf16 | bf16 | ≈ 16.6 GB | 24 GB+ | 27 |
| olmOCR-7B-0725-bf16 | bf16 | ≈ 16.6 GB | 24 GB+ | 24 |
| VL-Rethinker-7B-fp16 | fp16 | ≈ 16.6 GB | 24 GB+ | 18 |
| Jedi-7B-1080p-bf16 | bf16 | ≈ 16.6 GB | 24 GB+ | 16 |
| VisualQuality-R1-7B-bf16 | bf16 | ≈ 16.6 GB | 24 GB+ | 14 |
| olmOCR-2-7B-1025-mlx-8bit | 8-bit | — | — | 232K |
| olmOCR-2-7B-1025-mlx-4bit | 4-bit | — | — | 1.1K |
| olmOCR-2-7B-1025-4bit (this one) | 4-bit | — | — | 1.1K |
| olmOCR-2-7B-1025-8bit | 8-bit | — | — | 1.0K |
| olmOCR-7B-0725-8bit | 8-bit | — | — | 1.0K |
| olmOCR-2-7B-1025-6bit | 6-bit | — | — | 401 |
| olmOCR-2-7B-1025-5bit | 5-bit | — | — | 394 |
| VisualQuality-R1-7B-4bit | 4-bit | — | — | 59 |
| VisualQuality-R1-7B-8bit | 8-bit | — | — | 53 |
| Holo1-7B-6bit | 6-bit | — | — | 34 |
| olmOCR-7B-0725-4bit | 4-bit | — | — | 31 |
| Holo1-7B-4bit | 4-bit | — | — | 25 |
| VL-Rethinker-72B-4bit | 4-bit | — | — | 24 |
| Holo1-7B-8bit | 8-bit | — | — | 24 |
| VisualQuality-R1-7B-6bit | 6-bit | — | — | 23 |
| X-Reasoner-7B-8bit | 8-bit | — | — | 23 |
| Jedi-7B-1080p-6bit | 6-bit | — | — | 21 |
| olmOCR-7B-0725-6bit | 6-bit | — | — | 20 |
| VL-Rethinker-7B-6bit | 6-bit | — | — | 19 |
| VL-Rethinker-7B-8bit | 8-bit | — | — | 18 |
| VL-Rethinker-72B-fp16 | fp16 | — | — | 16 |
| VL-Rethinker-72B-6bit | 6-bit | — | — | 15 |
| VL-Rethinker-72B-8bit | 8-bit | — | — | 15 |
| Cosmos-Reason1-7B-6bit | 6-bit | — | — | 13 |
| Jedi-7B-1080p-8bit | 8-bit | — | — | 13 |