Text-to-Image
MLX
English
Chinese
image-generation
image-editing
quantized
4-bit precision
apple-silicon
minimax-h3-image
Instructions to use zimageapp/MiniMax-H3-Image-MLX-Q4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use zimageapp/MiniMax-H3-Image-MLX-Q4 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download zimageapp/MiniMax-H3-Image-MLX-Q4 --local-dir MiniMax-H3-Image-MLX-Q4
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Download benchmarks.json from zimageapp/MiniMax-H3-Image-MLX-Q4: direct link, hf CLI and curl.
- Browser
- Download file 2 kB
-
https://huggingface.co/zimageapp/MiniMax-H3-Image-MLX-Q4/resolve/main/benchmarks.json
- Command line
-
hf download hf://zimageapp/MiniMax-H3-Image-MLX-Q4/benchmarks.json
-
curl -L -o benchmarks.json https://huggingface.co/zimageapp/MiniMax-H3-Image-MLX-Q4/resolve/main/benchmarks.json
2 kB
| { | |
| "hardware": "Apple M5 Pro, 24 GB unified memory", | |
| "runtime": "Locraft native C++/Metal, BF16 compute, ANE off", | |
| "scope": "Single food prompt/seed; direct generations at different canvases. Engine process footprint is not whole-system unified memory. Timings include loading. These historical measurements used the tensor-identical aligned text encoder.", | |
| "runs": [ | |
| { | |
| "case": "q4-smoke", | |
| "width": 1024, | |
| "height": 768, | |
| "seed": 42, | |
| "steps": 6, | |
| "timing_s": { | |
| "denoise": 16.373842125, | |
| "dit_load": 0.16114175, | |
| "engine_total": 24.093915084, | |
| "text_encode": 6.819223416, | |
| "total": 24.1527019739151, | |
| "video_decode": 0.6763195 | |
| }, | |
| "process_peak_footprint_bytes": 3207236344 | |
| }, | |
| { | |
| "case": "q4-2k", | |
| "width": 2048, | |
| "height": 1536, | |
| "seed": 42, | |
| "steps": 6, | |
| "timing_s": { | |
| "denoise": 51.294036209, | |
| "dit_load": 0.143327042, | |
| "engine_total": 60.344243792, | |
| "text_encode": 6.8215445, | |
| "total": 60.5080269575119, | |
| "video_decode": 2.021023042 | |
| }, | |
| "process_peak_footprint_bytes": 3207858912 | |
| }, | |
| { | |
| "case": "q4-3k", | |
| "width": 3072, | |
| "height": 2304, | |
| "seed": 42, | |
| "steps": 6, | |
| "timing_s": { | |
| "denoise": 113.13924225, | |
| "dit_load": 0.115898375, | |
| "engine_total": 123.707646333, | |
| "text_encode": 6.318630542, | |
| "total": 124.02459502220154, | |
| "video_decode": 4.075760875 | |
| }, | |
| "process_peak_footprint_bytes": 3452046984 | |
| }, | |
| { | |
| "case": "q4-4k", | |
| "width": 4096, | |
| "height": 3072, | |
| "seed": 42, | |
| "steps": 6, | |
| "timing_s": { | |
| "denoise": 222.623776875, | |
| "dit_load": 0.123275833, | |
| "engine_total": 236.274012083, | |
| "text_encode": 6.447878875, | |
| "total": 236.813157081604, | |
| "video_decode": 7.005152375 | |
| }, | |
| "process_peak_footprint_bytes": 4393455768 | |
| } | |
| ] | |
| } | |