Metadata-Version: 2.3
Name: mlx-gen
Version: 0.18.12
Summary: Generative image and video model runtimes for MLX.
Keywords: flux,ai,ml,transformers,mlx,huggingface,apple-silicon,diffusers,qwen,qwen-image,seedvr2,z-image
Author: Filip Strand
Author-email: Filip Strand <strand.filip@gmail.com>
License: MIT License
         
         Copyright (c) 2026 Filip Strand
         
         Permission is hereby granted, free of charge, to any person obtaining a copy
         of this software and associated documentation files (the "Software"), to deal
         in the Software without restriction, including without limitation the rights
         to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
         copies of the Software, and to permit persons to whom the Software is
         furnished to do so, subject to the following conditions:
         
         The above copyright notice and this permission notice shall be included in all
         copies or substantial portions of the Software.
         
         THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
         IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
         FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
         AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
         LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
         OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
         SOFTWARE.
Classifier: Intended Audience :: Developers
Classifier: Operating System :: MacOS
Classifier: Operating System :: POSIX :: Linux
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Programming Language :: Python :: 3.14
Requires-Dist: av>=17.0.1
Requires-Dist: filelock>=3.20.1
Requires-Dist: fonttools>=4.60.2
Requires-Dist: huggingface-hub>=1.1.6,<2.0
Requires-Dist: hf-transfer>=0.1.9,<1.0
Requires-Dist: matplotlib>=3.9.2,<4.0
Requires-Dist: mlx>=0.27.0,<0.32.0 ; sys_platform == 'darwin'
Requires-Dist: mlx[cuda13]>=0.30.3,<0.32.0 ; sys_platform == 'linux'
Requires-Dist: numpy>=2.0.1,<3.0
Requires-Dist: opencv-python>=4.10.0,<5.0
Requires-Dist: piexif>=1.1.3,<2.0
Requires-Dist: pillow>=12.1.1
Requires-Dist: platformdirs>=4.0,<5.0
Requires-Dist: regex>=2024.11.6
Requires-Dist: requests>=2.32.4
Requires-Dist: safetensors>=0.4.4,<1.0
Requires-Dist: sentencepiece>=0.2.1,<1.0
Requires-Dist: tokenizers>=0.20.3 ; python_full_version >= '3.13'
Requires-Dist: toml>=0.10.2,<1.0
Requires-Dist: torch>=2.7.1,<3.0
Requires-Dist: torch>=2.8.0,<3.0 ; python_full_version >= '3.13'
Requires-Dist: tqdm>=4.66.5,<5.0
Requires-Dist: transformers>=5.0.0,<6.0
Requires-Dist: twine>=6.1.0,<7.0
Requires-Dist: urllib3>=2.6.0
Requires-Dist: protobuf>=4.25.0,<8.0
Requires-Dist: matplotlib>3.10,<4.0 ; extra == 'dev'
Requires-Dist: pytest>=8.3.0,<9.0 ; extra == 'dev'
Requires-Dist: pytest-timer>=1.0,<2.0 ; extra == 'dev'
Requires-Dist: mlx==0.31.0 ; sys_platform == 'darwin' and extra == 'dev'
Requires-Dist: mlx[cuda13]==0.30.3 ; sys_platform == 'linux' and extra == 'dev'
Maintainer: AbstractVision
Requires-Python: >=3.10
Project-URL: homepage, https://github.com/lpalbou/mlx-gen
Project-URL: upstream, https://github.com/filipstrand/mflux
Provides-Extra: dev
Description-Content-Type: text/markdown

# MLX-Gen

[![mlx-gen](https://img.shields.io/pypi/v/mlx-gen?label=mlx-gen&logo=pypi&logoColor=white)](https://pypi.org/project/mlx-gen/)
[![MLX](https://img.shields.io/pypi/v/mlx?label=MLX&logo=pypi&logoColor=white)](https://pypi.org/project/mlx/)
[![CI](https://github.com/lpalbou/mlx-gen/actions/workflows/tests.yml/badge.svg)](https://github.com/lpalbou/mlx-gen/actions/workflows/tests.yml)

MLX-Gen is a local image and video generation runtime for Apple Silicon and MLX. It exposes
`mlxgen` for text-to-image, image-to-image, text-to-video, image-to-video, model download, model
preparation, quantized local folders, and application progress callbacks. It also includes the
dedicated `mflux-upscale-seedvr2` command for SeedVR2 image super-resolution.

> [!IMPORTANT]
> MLX-Gen started as a fork of [mflux](https://github.com/filipstrand/mflux). Most credit for the
> current codebase goes to Filip Strand and the original mflux contributors. This project keeps
> that attribution visible while publishing independently as `mlx-gen` and evolving the `mlxgen`
> command surface for current Apple Silicon workflows.

![MLX-Gen workflow example](https://raw.githubusercontent.com/lpalbou/mlx-gen/main/docs/assets/examples/spaceship-snow/mlx-gen-example.png)

## What It Does

MLX-Gen runs supported Hugging Face and prepared MLX-Gen model folders without starting network
downloads during generation. You explicitly download or prepare models first, then generation is a
cache-only operation suitable for desktop apps, workflow engines, and long-running local jobs.

The main capabilities are:

- text-to-image generation with Qwen Image, FLUX.2 Klein, Z-Image, ERNIE Image Turbo, Bonsai Image,
  FIBO, and related prepared folders;
- image-to-image modes, including latent img2img, instruction/reference edits, and multi-reference
  edits where the selected model supports them;
- Wan2.2 text-to-video and image-to-video, including TI2V-5B BF16/q8 packages plus A14B
  T2V/I2V prepared BF16 and mixed q8/BF16 packages; Wan I2V resolves output size from the source
  image aspect ratio so inputs are not stretched into a mismatched canvas;
- SeedVR2 image super-resolution through `mflux-upscale-seedvr2`, with shortest-edge target sizing
  or explicit scale factors such as `2x` and `3x`;
- explicit `download` and `prepare` workflows for reproducible local model folders;
- JSON model capability inspection before starting a heavy run;
- shared progress events for applications embedding MLX-Gen.

Use `mlxgen capabilities --model ...` before long image-edit runs. Capability output describes the
available route; validation reports and contact sheets describe whether an exact source handle or
prepared package passed a visual release gate. Release evidence should use true handles such as
`briaai/Fibo-Edit` or `AbstractFramework/flux.2-klein-9b-8bit`, not short aliases.

## Install

Install with `uv`:

```sh
uv tool install --upgrade mlx-gen
```

Or install into an environment:

```sh
python -m pip install -U mlx-gen
```

Check the command surface:

```sh
mlxgen --help
```

## First Commands

Download model files explicitly:

```sh
mlxgen download --model AbstractFramework/flux.2-klein-9b-8bit
```

Generate an image:

```sh
mlxgen generate \
  --model AbstractFramework/flux.2-klein-9b-8bit \
  --prompt "A cinematic wide shot of a compact sci-fi spaceship resting in deep snow on a frozen alien planet" \
  --width 768 \
  --height 432 \
  --steps 24 \
  --guidance 1.0 \
  --seed 6107 \
  --output spaceship.png
```

Upscale an image with SeedVR2:

```sh
mflux-upscale-seedvr2 \
  --image-path input.png \
  --resolution 2x \
  --quantize 8 \
  --softness 0.25 \
  --metadata \
  --output input_2x.png
```

For SeedVR2, an integer `--resolution` is the target shorter edge while values such as `2x` and
`3x` are scale factors. Both modes preserve the source aspect ratio. Use `--softness 0.25` to
`0.5` when the source has visible grain in smooth areas; keep the default no-tiling path for best
quality, and add `--vae-tiling` only for very large memory-bound upscales. See
[docs/upscaling.md](docs/upscaling.md) for a reproducible 5x SeedVR2 comparison.

Inspect model capabilities before a run:

```sh
mlxgen capabilities --model AbstractFramework/flux.2-klein-9b-8bit
```

Capabilities are route contracts: they show which tasks, I2I modes, image counts, and options the
selected model can dispatch. For release QA evidence on exact packages, use:

```sh
mlxgen validation --model AbstractFramework/qwen-image-edit-2509-8bit
```

Create a reusable local prepared folder:

```sh
mlxgen prepare \
  --model Qwen/Qwen-Image \
  --path ./models/qwen-image-8bit \
  --quantize 8
```

`mlxgen generate` does not download missing files. If something is not cached, MLX-Gen raises a
clear `DownloadRequiredError` with the command to run. A complete prepared folder at
`./models/<repo-name>` can also satisfy a matching Hugging Face handle such as
`AbstractFramework/qwen-image-edit-2511-8bit`.

## Reproducible Example

The docs include a complete model-backed spaceship workflow:

- T2I: generate a spaceship in the snow.
- I2I edit: turn it into a pencil sketch.
- I2I edit: crash the same spaceship in the snow.
- I2I multi-reference: combine the crash layout and pencil-sketch style.
- T2V A14B: generate a spaceship taking off from a snow planet.
- I2V A14B: animate the generated spaceship taking off from the source image.

See [docs/examples/spaceship-snow.md](docs/examples/spaceship-snow.md) for the exact commands and
included assets.

![Spaceship mode contact sheet](https://raw.githubusercontent.com/lpalbou/mlx-gen/main/docs/assets/examples/spaceship-snow/spaceship_modes_real_generation_contact_sheet.png)

For current image-edit validation evidence across Qwen Image Edit, Qwen EditPlus, FLUX.2 Klein,
and latent I2I models, see
[docs/edit-capabilities.md](docs/edit-capabilities.md).

## Published Models

Prepared MLX-Gen model folders are published under the
[AbstractFramework organization on Hugging Face](https://huggingface.co/AbstractFramework). Current
published examples include:

- `AbstractFramework/flux.2-klein-4b-4bit`
- `AbstractFramework/flux.2-klein-4b-8bit`
- `AbstractFramework/flux.2-klein-9b-4bit`
- `AbstractFramework/flux.2-klein-9b-8bit`
- `AbstractFramework/qwen-image-2512-4bit`
- `AbstractFramework/qwen-image-2512-8bit`
- `AbstractFramework/qwen-image-edit-2509-8bit`
- `AbstractFramework/qwen-image-edit-2511-4bit`
- `AbstractFramework/qwen-image-edit-2511-8bit`
- `AbstractFramework/z-image-turbo-4bit`
- `AbstractFramework/z-image-turbo-8bit`
- `AbstractFramework/ernie-image-turbo-4bit`
- `AbstractFramework/ernie-image-turbo-8bit`
- `AbstractFramework/wan2.2-ti2v-5b-diffusers-bf16`
- `AbstractFramework/wan2.2-ti2v-5b-diffusers-8bit`
- `AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16`
- `AbstractFramework/wan2.2-t2v-a14b-diffusers-8bit`
- `AbstractFramework/wan2.2-i2v-a14b-diffusers-bf16`
- `AbstractFramework/wan2.2-i2v-a14b-diffusers-8bit`

Use `mlxgen download --model <repo-id>` to cache a published model, or pass the repository id
directly to `mlxgen generate` after it is cached. See
[docs/quantization.md](docs/quantization.md) for the complete current package matrix with source
sizes, prepared package sizes, task coverage, and quantization notes.

For Wan2.2 TI2V-5B, the published BF16 prepared package is 21.2 GiB versus 31.9 GiB for the
upstream source snapshot. It is mainly a smaller reusable source-equivalent package because
MLX-Gen already loads Wan transformer/VAE weights at BF16 runtime precision. The published q8
package is 16.9 GiB. In the documented 1280x704 benchmark profile, q8 reduced logical model
footprint and MLX allocator peak but did not reduce full-process physical peak memory. Wan TI2V-5B
q4 or mixed q4/q8 is not published as a supported package. See the exact benchmark profile in
[docs/quantization.md](docs/quantization.md).

## Wan A14B Measurements

Wan A14B was measured on an Apple M5 Max with 128 GB unified memory. The published-card benchmark
uses small, repeatable low-RAM runs and records full-process Darwin physical footprint, RSS, MLX
allocator peak, and generation time. Use these values for the listed profiles; memory and runtime
scale with resolution, frame count, step count, cache settings, and image-to-video conditioning.

| Model | Package | Disk | Physical Peak | Max RSS | MLX Peak | Time | Profile |
| --- | --- | ---: | ---: | ---: | ---: | ---: | --- |
| Wan2.2 T2V-A14B | BF16 | 64.1 GiB | 33.0 GiB | 31.8 GiB | 27.7 GiB | 152.7 s | 384x224, 33 frames, 12 steps, 8 fps |
| Wan2.2 T2V-A14B | mixed q8/BF16 | 39.5 GiB | 20.7 GiB | 19.5 GiB | 15.5 GiB | 154.8 s | 384x224, 33 frames, 12 steps, 8 fps |
| Wan2.2 I2V-A14B | BF16 | 64.1 GiB | 33.7 GiB | 31.8 GiB | 28.2 GiB | 228.2 s | 384x384, 33 frames, 12 steps, 8 fps |
| Wan2.2 I2V-A14B | mixed q8/BF16 | 39.5 GiB | 21.5 GiB | 19.6 GiB | 15.9 GiB | 242.2 s | 384x384, 33 frames, 12 steps, 8 fps |

In these runs, mixed q8/BF16 reduces disk usage by about 38% versus prepared BF16 folders and
reduces full-process physical peak memory by about 36-37%. It is not documented as a speed
improvement. See
[docs/quantization.md](docs/quantization.md) for model-family quantization details and metrics JSON.
The 0.18.11 release also validates the published A14B q8 T2V/I2V handles on a 41-frame,
15-step, 480x240-target profile with saved MP4/contact-sheet evidence in the quantization docs.

## Ecosystem

MLX-Gen is used as the local Apple Silicon generation backend for:

- [AbstractVision](https://github.com/lpalbou/abstractvision), the vision/generation layer of the
  AbstractFramework ecosystem;
- [AbstractFramework](https://github.com/lpalbou/abstractframework), the broader framework for
  local agentic and generative workflows;
- [AbstractFlow](https://github.com/lpalbou/abstractflow), a visual orchestration layer that can
  compose generative capabilities with persistent agentic tasks.

MLX-Gen remains useful as a standalone CLI package, but its cache-only runtime behavior, capability
inspection, prepared model folders, and progress callbacks are designed so applications can embed it
without surprise network transfers or ambiguous model routing.

## Documentation

- [Getting started](docs/getting-started.md): installation, first runs, SeedVR2 upscaling, and Wan video.
- [API and CLI](docs/api.md): command surface, router behavior, image-to-image modes, SeedVR2 sizing, Wan video sizes, capabilities, and Python entry points.
- [Example workflow](docs/examples/spaceship-snow.md): reproducible image and video commands.
- [Image upscaling](docs/upscaling.md): SeedVR2 sizing, quality controls, and a 5x source/output comparison.
- [Image edit capabilities](docs/edit-capabilities.md): visual edit-validation contact sheets, exact model/package status, and command logs.
- [Model management](docs/model-management.md): download, prepare, cache-only runtime policy.
- [Quantization](docs/quantization.md): q8/q4/BF16 policies and measurements.
- [Python integration](docs/python-integration.md): embedding, progress callbacks, and AbstractVision notes.
- [FAQ](docs/faq.md): recurring questions, image-to-image mode selection, SeedVR2 sizing, Qwen edit variants, negative prompts, outpaint/reframe status, Wan resolutions, and usage limits.
- [Troubleshooting](docs/troubleshooting.md): common setup and runtime failures.
- [Acknowledgements](ACKNOWLEDGEMENTS.md): upstream mflux and model-community credits.

## License

MLX-Gen is MIT licensed. Model weights remain governed by their original licenses and access terms.
