This release is a pre-release and may not be stable for production use.
mlx-beam-vision
The towers that let mlx-beam see. Install
it next to the engine (pip install mlx-beam[vision], or this package by
name) and a checkpoint that ships a vision tower serves images: an
image_url part with a data: URL in a chat request, the checkpoint's own
processor renders the template, the tower encodes the image, and the
engine's prefill takes the image's features in place of the placeholder
tokens - beside text-only requests in the same batch, with the prefix cache
keyed on the image's digest. Without this package the engine answers image
input with a 400 that names the vision extra.
Towers vendored from mlx-vlm (MIT,
see mlx_beam_vision/_vendor/VENDORED.md):
| Family | Models | Notes |
|---|---|---|
qwen3_vl |
Qwen3-VL, Qwen3.5, Qwen3.8 | DeepStack applied: the tower's intermediate features are added after the text model's first layers; the text model's multimodal positions (MRoPE: time, height, width per image token) are built from the processor's grids and applied through prefill, cache and decoding |
mistral3 |
Mistral Small 3.x (Pixtral tower) | the image's rows are separate spans ([IMG_BREAK] between them) |
gemma4 |
Gemma 4 E2B, E4B | image only; audio is not served yet. The 12B (gemma4_unified, encoder-free, its image tokens attending bidirectionally in the text model) is refused with that reason until the text trunk takes such a mask |
muse_glimmer |
Muse Glimmer | the projector's norm matches the text model's input norm |
granite4_vision |
Granite Vision 4.1 | AnyRes tiles, window Q-Former projectors; nothing at the embedding, every projector's features added ahead of its text layer |
Not yet: audio and video input, quantized towers (the towers load at model precision; a package that ships a quantized tower is refused with a message). An image above 32 megapixels is refused before it is decoded.
The package installs torch and torchvision: the processors transformers ships for these families (the template rendering and the image-to-patches step) are written on them. The towers themselves run on MLX; torch does no inference here.
Every family is exercised in the tests on a tiny random tower against a direct forward; the real towers and processors are validated on Apple silicon with the checkpoints themselves.
/health.vision reports the family, the tower, the processor and the
feature cache (encoder outputs by image digest, bounded in bytes).
1 maintainer
Metadata
1 maintainer
Release files for mlx-beam-vision 0.1.0a8
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| mlx_beam_vision-0.1.0a8.tar.gz | 52.8 kB | With provenance | Details for mlx_beam_vision-0.1.0a8.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| mlx_beam_vision-0.1.0a8-py3-none-any.whl | Python 3 | none | any | Details for mlx_beam_vision-0.1.0a8-py3-none-any.whl |
Total release size: 114.1 kB
Release files / mlx_beam_vision-0.1.0a8.tar.gz
| Download URL | mlx_beam_vision-0.1.0a8.tar.gz |
|---|---|
| Size | 52.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
bb792c85f3c3a564c6718cdea438cec5b563e481eb5536095de92fd5605a05cf
|
|
BLAKE2b-256 checksum How to use checksums |
15f6549ba8d6b401255607f4c901c8b0a4c223df303e65601adb47875c53629e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 26, 2026.
Transparency logRelease files / mlx_beam_vision-0.1.0a8-py3-none-any.whl
| Download URL | mlx_beam_vision-0.1.0a8-py3-none-any.whl |
|---|---|
| Size | 61.4 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
8cc5a25efe6eca423f0d474c6554b87235b8294d8969612b37e1771526956b36
|
|
BLAKE2b-256 checksum How to use checksums |
2cefd9e94fae78c89f3ab1b3b1c97a27071a66de621635c78829c9d156072922
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 26, 2026.
Transparency logSecurity for mlx-beam-vision 0.1.0a8
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for mlx_beam_vision-0.1.0a8.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 26, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
