FLUX
3 aliases · FLUX.2 klein 4B (4-bit and full precision) and FLUX.1-schnell 12B. rapid-mlx serve flux2-klein-4b is the fast default and runs on a 12 GB Mac.
Black Forest Labs' FLUX.2 klein in its 4B form, converted to mflux 4-bit: a 4.6 GB download that renders in 4 steps on a 12 GB Mac and also takes image edits on /v1/images/edits. The full-precision build (flux2-klein-4b-bf16) wants a 32 GB Mac; FLUX.1-schnell (flux-schnell, 12B, 4-bit) a 16 GB one. Same OpenAI-compatible /v1/images/generations endpoint and desktop Images tab support as the rest of the lane.
- family
- FLUX (Black Forest Labs)
- aliases
- 3
- install
- pip install 'rapid-mlx[image]'
- OpenAI base URL
- http://localhost:8000/v1
Usage
rapid-mlx serve flux2-klein-4b
curl http://localhost:8000/v1/images/generations \
-H 'Content-Type: application/json' \
-d '{"model":"flux2-klein-4b","prompt":"a paper crane on wet slate, studio light","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 flux2-klein-4bFLUX.2 klein 4B · 4.6 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 |
|---|---|---|---|
| flux2-klein-4b | Runpod/FLUX.2-klein-4B-mflux-4bit | 12 GB | The fast default — 4 steps, generate and edit. |
| flux2-klein-4b-bf16 | mflux-community/flux2-klein-4b-mflux-bf16 | 32 GB | Full precision; select it with --image-weight-precision bf16 or by alias. |
| flux-schnell | mflux-community/flux-1-schnell-mflux-q4 | 16 GB | FLUX.1-schnell 12B, 4-bit, 4 steps. |
Frequently asked questions
Can I run FLUX locally on a Mac?
Yes — flux2-klein-4b renders entirely on Apple Silicon through POST /v1/images/generations. 4.6 GB download, 12 GB Mac.
Which Mac do I need for FLUX.2 klein?
12 GB of unified memory for the 4-bit flux2-klein-4b; 32 GB for the full-precision build.