Qwen Image
4 aliases · Qwen-Image, Qwen-Image 2.1 and Qwen-Image-Edit. rapid-mlx serve qwen-image-2.1 generates and edits on an 8 GB+ Mac; the q6 Qwen-Image wants 64 GB.
Alibaba's image models, the strongest text-in-image of the lane. qwen-image-2.1 is an mflux 4-bit build that generates and edits (img2img) from an 8 GB Mac; qwen-image-2.1-bf16 is its full-precision twin (32 GB). qwen-image is the original model in mflux q6 (31 GB of weights, 64 GB Mac), and qwen-image-edit is the instruction-edit model (96 GB Mac, /v1/images/edits only). Same OpenAI-compatible /v1/images/generations endpoint and desktop Images tab support as the rest of the lane.
- family
- Qwen Image (Alibaba)
- aliases
- 4
- install
- pip install rapid-mlx==0.16.0
- OpenAI base URL
- http://localhost:8000/v1
Usage
rapid-mlx serve qwen-image
curl http://localhost:8000/v1/images/generations \
-H 'Content-Type: application/json' \
-d '{"model":"qwen-image","prompt":"an ink-wash mountain village at dawn","size":"1024x1024"}'
Download
Mirrored on the rapid-mlx CDN — with automatic mid-pull fallback to Hugging Face if a mirror file slows down. One command, no account:
rapid-mlx pull qwen-imageQwen Image (mflux q6) · 31 GB · mirrored ✓ · live status
Weights land in the standard Hugging Face cache, and rapid-mlx serve pulls automatically on first use.
Aliases
| alias | hf repo | min RAM | notes |
|---|---|---|---|
| qwen-image | mflux-community/qwen-image-mflux-q6 | 64 GB | The original Qwen-Image, mflux q6. |
| qwen-image-2.1 | mlx-community/Qwen-Image-2.1-mflux-q4 | 8 GB | Generate and edit, mflux 4-bit. |
| qwen-image-2.1-bf16 | Qwen/Qwen-Image-2.1 | 32 GB | Full precision; pulls from Hugging Face. |
| qwen-image-edit | OsaurusAI/Qwen-Image-Edit-mflux-q8 | 96 GB | Instruction edits only (/v1/images/edits); pulls from Hugging Face. |
Frequently asked questions
Can I run Qwen Image locally on a Mac?
Yes — qwen-image renders entirely on Apple Silicon through POST /v1/images/generations. 31 GB download, 64 GB Mac.
How much RAM does Qwen Image need?
8 GB for qwen-image-2.1, 32 GB for its full-precision build, 64 GB for qwen-image and 96 GB for qwen-image-edit.