High-performance MLX-native nodes for Nodetool on Apple Silicon. This package wraps the community MLX implementations of Whisper, Kokoro/Sesame TTS, MFlux FLUX.1 image generation, and Stability AI's Stable Audio 3 so you can run state-of-the-art audio and vision workflows locally on macOS.
- Local-first – keep data on-device by running speech, TTS, and image models without cloud calls
- Optimised for Apple Silicon – uses MLX kernels and quantized checkpoints to achieve strong throughput on M-series chips
- Drop-in nodes – integrates seamlessly with the Nodetool graph editor and
nodetool-coreruntime
All nodes live under src/nodetool/nodes/mlx. Audio nodes wrap the
mlx-audio library.
KokoroTTS– fast multilingual TTS with 54 voice presetsSesameTTS– CSM voice cloning from a reference clipSparkTTS– controllable speed / pitch / gender presetsQwen3TTS– multilingual TTS with speaker voices and voice designKittenTTS– compact, edge-friendly English voicesDiaTTS– dialogue TTS with[S1]/[S2]speaker tagsOuteTTS– efficient multilingual TTS with optional cloningOmniVoiceTTS– zero-shot multilingual (646+ languages)MeloTTS– lightweight VITS2 English accentsVoxtralTTS– Mistral's multilingual TTS with voice presetsChatterboxTTS– expressive TTS with exaggeration control and cloningHiggsAudioTTS– conversational TTS with zero-shot cloningLongCatAudioTTS– diffusion TTS with zero-shot cloningMLXTextToSpeech– generic node that runs anymlx-audioTTS repo id
Whisper– MLX Whisper transcription with optional word timestampsParakeet– NVIDIA Parakeet high-accuracy multilingual ASRQwen3ASR– Alibaba's multilingual ASR with long-form chunkingQwen3ForcedAligner– word-level timestamp alignment for a known transcriptMLXSpeechToText– generic node that runs anymlx-audioSTT repo id
DeepFilterNet– real-time noise suppression (v1/v2/v3) at 48 kHzMossFormer2– high-quality 48 kHz speech enhancement
MFluxand theMFlux*family – FLUX.1 / Qwen-Image / Z-Image / FIBO generation and editing via the MFlux project (supports quantized models)
MLXVisionLanguage– image captioning, visual Q&A, and OCR via MLX vision-language models (Qwen3-VL, Gemma 4)
StableAudio3– text-to-audio music & sound effects with Stable Audio 3 (44.1 kHz stereo)StableAudio3AudioToAudio– prompt-guided variations of an input clipStableAudio3Inpaint– regenerate a time range inside an audio clip
ACEStepMusicGeneration– local text-to-music generation with ACE-Step 1.5 (prompt + lyrics → full songs)ACEStepSongPlanner– use the ACE-Step 5Hz language model to turn an idea into a caption, lyrics and musical metadata
TextGeneration– local LLM text generation viamlx-lm
The Stable Audio 3 nodes run Stability AI's optimized MLX implementation (no PyTorch at
runtime), vendored under nodetool.mlx.stable_audio_3 (MIT licensed — see that folder's
LICENSE and NOTICE.md). Three DiT variants are available via the model field:
sm-music(50M) – fast music generationsm-sfx(50M) – sound effectsmedium(1.4B) – higher-fidelity music
Weights are pulled on demand from the Hugging Face repo
stabilityai/stable-audio-3-optimized
(only the MLX/* files) and cached locally; you can also pre-download them from the
Models Manager.
The ACE-Step nodes wrap the official ACE-Step 1.5
project, which uses MLX for the diffusion transformer, VAE and language model on
Apple Silicon. ACE-Step 1.5 is not distributed on PyPI (the ace-step package
there is the older 1.0 release), so it must be installed separately:
git clone https://github.com/ace-step/ACE-Step-1.5.git
cd ACE-Step-1.5
uv sync # installs the macOS/MLX dependenciesInstall it into the same environment as nodetool-mlx. If the package is missing,
the nodes raise a clear error pointing back to the repository.
Checkpoints are downloaded automatically from HuggingFace on first use into
~/.cache/nodetool/acestep/checkpoints (override with the ACESTEP_CHECKPOINTS_DIR
environment variable). The main bundle ACE-Step/Ace-Step1.5 includes the VAE, the
turbo DiT and the 1.7B planner LM; additional DiT and LM checkpoints are listed as
recommended models on each node.
- macOS 14+ on Apple Silicon (MLX currently supports Apple hardware only)
- Python 3.11
- nodetool-core v0.6.0+
- Required MLX checkpoints managed via the Nodetool Models Manager (see Managing Models)
- Open Nodetool → Tools ▸ Packages
- Install the
nodetool-mlxpack from the package registry - Nodetool will handle dependencies and expose the MLX nodes in the graph editor once installed
git clone https://github.com/nodetool-ai/nodetool-mlx.git
cd nodetool-mlx
uv pip install -e .
uv pip install -r requirements-dev.txtIf you prefer Poetry or pip, install the project the same way—just ensure dependencies are resolved against Python 3.11.
All MLX nodes rely on locally cached checkpoints. The recommended way to download and update them is through the Models Manager built into Nodetool:
- Open Nodetool → Menu ▸ Models
- Select the
mlxtab to view the recommended checkpoints for each node - Click Download for the models you plan to use; Nodetool stores them in the Hugging Face cache automatically
- The UI will keep track of model availability and prompt you when updates are available
Advanced users can still seed the Hugging Face cache manually, but using the UI integration ensures consistent paths and avoids missing-model errors in workflows.
- Install
nodetool-coreand this package in the same environment - Run
nodetool-pkg scan --write --enrichto generate package metadata - Build workflows in the Nodetool UI using the
mlxnodes
Run tests and lint checks before submitting PRs:
pytest -q
ruff check .
black --check .Most of the suite is written to run on any platform: the MLX runtimes are
imported lazily and the node logic is exercised against mocks. The handful of
tests that need mflux or mlx-audio skip themselves when those Apple Silicon
packages are unavailable, so a Linux run reports skips rather than failures.
If you change a node's get_recommended_models(), regenerate the committed
package metadata:
nodetool-pkg scan --write --enrichtests/test_package_metadata.py fails when the two drift apart — the Nodetool
UI reads that file to populate the model picker, so a stale entry becomes a
model users cannot select or a download offer for a repository that no longer
exists.
.github/workflows/ci.yml runs on every pull request:
| Job | Runner | What it covers |
|---|---|---|
| Lint and format | ubuntu | ruff check . and black --check . |
| Test (linux) | ubuntu, Python 3.11 + 3.12 | the platform-independent suite, installed with --no-deps |
| Test (macOS) | macos-14 (Apple Silicon) | the full suite against the real MLX stack |
| Build wheel | ubuntu | python -m build plus twine check |
Please open issues or pull requests for bug fixes, new MLX models, or performance improvements. Contributions are welcome!