Joint audio+video: one model generates the frames and their soundtrack together.
Text-to-video, image-to-video (first-frame conditioning) and audio-to-video, plus
the two-stage pipelines, all served natively — no Python.
36 GB on disk, down from 110 GB. The upstream conversion ships bf16 only; this
is the same pack at the 4-bit width mlx-serve's LTX engine reads, and it is
self-contained — the text encoder is inside it, so nothing else downloads.
⚠️ License — read before use
These weights are Derivatives of LTX-2.5 under the LTX-2.x Community
License Agreement. A complete copy ships here as LICENSE.md, and
the Acceptable Use Policy it incorporates by reference is snapshotted as
ltx-acceptable-use-policy-snapshot-2026-08-12.pdf
(the version in effect at your time of use governs — check
Lightricks' current AUP).
Transfer notice (Agreement §3.5). Your use of these weights is subject to the
LTX-2.x Community License Agreement. If you (aggregated across entities under
common control) have annual revenues of US $10,000,000 or more, you are a
"Commercial Entity" under the Agreement and must obtain a paid license from
Lightricks before any use other than the Agreement's non-commercial-purpose
carve-outs (testing, evaluation, non-commercial R&D in non-production
environments). Sub-threshold commercial and production use is royalty-free under
the Agreement's terms.
Further obligations that travel with these weights include (not exhaustive — read
the license): machine-generated content disclosure (Attachment A §5), no removal
or circumvention of any transparency/provenance features (§6), EU AI Act /
CA AI Transparency Act responsibilities for providers/deployers (§6), and the
Attachment A acceptable-use terms.
Modification notice (Agreement §3.3)
Modified from mlx-community/ltx-2.5-mlx. The change is quantization only — no
training, no fine-tuning, no architectural change:
Both DiT variants: affine 4-bit, group size 64, applied to the 34 linear
projections in each of the 48 transformer blocks (1632 weights per variant) —
the six attention modules' to_q/to_k/to_v/to_out/to_gate_logits and
the two feed-forwards' proj_in/proj_out. 38.0 GB → 11.3 GB each.
Text encoder (gemma4-12b-ltx-v1/): affine 4-bit, group size 64, on every 2-D
projection plus the embedding table (329 weights). 23.8 GB → 6.7 GB.
Kept at their shipped dtype: every adaLN table, patchify_proj/proj_out,
the q/k norms, keyframes_abs_pos_embedding, and all of the connector, both
video VAEs, the audio VAE, the vocoder and both latent upscalers — copied
byte-for-byte.
That split is the same recipe the established 2.3 MLX packs use
(only_transformer_blocks), which is what mlx-serve's loader expects: it treats a
tensor as quantized iff a sibling <name>.scales exists.
Conversion script: scripts/quantize_ltx25.py.
The embedded LTX-2.x license text in upstream file metadata is preserved.
1curl -X POST http://127.0.0.1:11434/v1/video/generations \2 -H 'Content-Type: application/json'\3 -d '{"prompt":"a red fox trots across fresh snow in a pine forest at golden hour",
4 "num_frames":97,"height":480,"width":704,"steps":8,"seed":42}'
Or pick LTX-Video 2.5 in the mlx-serve app's Video window.
Measured
M4 Max (128 GB): 704×480, 97 frames @ 24 fps, 8 steps — 2m17s end to end,
including the text encode and the VAE decode, with the generated soundtrack.
