--- title: ltx-2.5-mlx-q8 canonical_url: "https://www.modelscope.cn/models/mlx-community/ltx-2.5-mlx-q8" md_url: "https://www.modelscope.cn/models/mlx-community/ltx-2.5-mlx-q8.md" repository: mlx-community/ltx-2.5-mlx-q8 last_updated: 2026-10-16 license: other pipeline_tag: text-to-video-synthesis tasks: - text-to-video-synthesis model_type: - AudioVideo base_model: - Lightricks/LTX-2.5 base_model_relation: finetune parameters: 9.2B tensor_type: - BF16 - U32 library_name: - mlx - safetensors - pytorch frameworks: - pytorch downloads: 917 stars: 0 tags: - mlx - ltx-video - ltx-2.5 - text-to-video - image-to-video - audio-video - quantized --- # ltx-2.5-mlx-q8 > ltx-2.5-mlx-q8 - mlx-community 在 ModelScope 开源的模型。LTX-2.5 — int8 text encoder (MLX) mlx-community/ltx-2.5-mlx-q8 是 ModelScope 魔搭社区上的 9.2B 参数text-to-video-synthesis模型，采用 other 许可，基于 Lightricks/LTX-2.5 构建。 - **Repository**: mlx-community/ltx-2.5-mlx-q8 - **License**: other - **Tasks**: text-to-video-synthesis - **Parameters**: 9.2B - **Base model**: Lightricks/LTX-2.5 - **Tags**: mlx, ltx-video, ltx-2.5, text-to-video, image-to-video, audio-video, quantized - **Downloads**: 917 - **Stars**: 0 - **Last updated**: 2026-10-16 Source: https://www.modelscope.cn/models/mlx-community/ltx-2.5-mlx-q8 --- # LTX-2.5 — int8 text encoder (MLX) A complete LTX-2.5 component tree whose **only difference from [`mlx-community/ltx-2.5-mlx`](https://huggingface.co/mlx-community/ltx-2.5-mlx) is the Gemma-4 text encoder, quantized to int8** (13.6 GB vs 23.8 GB). Everything else is byte-identical to the base repo and is duplicated here only so this tree resolves standalone. ⚠️ **This repo does NOT contain a quantized DiT.** The transformer is a separate axis: see [`mlx-community/ltx-2.5-mlx-ditq8`](https://huggingface.co/mlx-community/ltx-2.5-mlx-ditq8). The two are independent — you can mix an int8 encoder with a bf16 or int8 DiT. ## Why it exists — it is load-bearing on small machines, not a nicety With the DiT streamed, LTX-2.5's peak memory is set by a **geometry-independent ~24.8 GB floor** that turns out to be the bf16 encoder (measured: a 256×256×9 clip and a 512×288×121 clip peak within 0.01 GB of each other — a decode-bound peak would scale with geometry; this does not). That floor is over the budget of both 24 GB and 32 GB machines. Swapping in this int8 encoder takes the whole-run peak to **14.6 GB / 15.4 GB** respectively. **Neither lever works alone**: resident is DiT-bound, so the encoder swap is invisible without streaming, and streaming alone leaves the encoder floor. ## Recipe `int8`, **group 64**, with **`embed_tokens` deliberately EXCLUDED** (kept bf16). Hidden state 00 *is* `embedTokens(ids) · embedScale` — one of the 49 states the connector consumes directly, and the input to all 48 layers. Quantizing it moves that state on its own; `state 00 == 1.000000` exactly is the fingerprint that it stayed bf16. ## Fidelity Connector output at **valid token positions** — the number that decides a text-encoder quant — reads **0.999820** against a **0.999879** bf16 floor. ⚠️ Read that metric, not a whole-array cosine. With a short prompt most of the 1024 positions are learned registers, identical across every encoder arm, which dilutes a whole-array figure 3–30×: a deliberately-poisoned arm scored 0.994316 whole-array but 0.829329 on its real tokens. ⚠️ **The sample MOVES.** Output is not bit-identical to bf16 (mean −0.1421/σ 0.4253 vs −0.1274/0.4222) — a different but equally valid video, because changing the encoder changes the conditioning and therefore the trajectory. Judged acceptable in a **blind** 4-pair perceptual A/B across prompts including a close-up face, at both 704×512 and a clamped 512×288 one-stage geometry. Comparing the arms with SSIM/PSNR would be a category error. ## Licence LTX-2 Community License — see `LICENSE.md` and the acceptable-use policy snapshot. Derivative of Lightricks LTX-2.5, carrying its terms including the Attachment A revenue gate and non-compete. Redistribution permitted under §3 with those terms propagated. ## `vae_decoder_pruna.safetensors` — PrunaAI's lean video decoder (optional) **What it is.** [PrunaAI/PrunaVAED](https://huggingface.co/PrunaAI/PrunaVAED) (revision `4baacd7ef66a6131439542c1f05872afe042e128`, last modified 2026-07-29), a pruned, distillation-finetuned drop-in replacement for the LTX-2 video VAE *decoder*, converted on 2026-07-31 from its diffusers layout to this repo's `ltx-core` flat decoder dialect with [`scripts/convert_pruna_vae_decoder.py`](https://github.com/xocialize/ltx-2-mlx/blob/ltx-2.5/scripts/convert_pruna_vae_decoder.py). The conversion is a key mapping plus the PyTorch→MLX conv transposition; no tensor values are changed. Only the decoder is produced: the encoder, the latent format and the per-channel statistics are the stock ones, and the video VAE is byte-identical across LTX-2.3 and LTX-2.5, so one file serves both. It is not a pure width prune — two blocks carry a channel-adapter ResBlock (`conv_in` 512→384 and 384→256, with an affine `norm3`) that the stock decoder lacks; the MLX decoder derives every width from the weights and enables those adapters on sight. **Why it is here.** Decode is roughly 2× faster and the decode activation peak materially smaller, at a decode that is near-identical but not bit-exact. Measured on a 24 GB M5 Pro at the compact24 tier (receipt AB-R-0561, `ltx-2-mlx-swift`): −15.9 % wall-clock per clip at the tier's own envelope with the peak unchanged, and where decode binds (704×512×121) a 21.1 → 14.5 GB peak that turns an over-budget, swapping run into one within budget. The Swift port selects it with `LTX2Configuration.vaeDecoderPath` or `LTX_VAE_DECODER=pruna`, and makes it the default decoder on the low tiers (compact24, balanced32) once promoted. The port reads its components from this int8-encoder sibling tree whenever the int8 text encoder is selected (the shipping low-tier configuration), which is why the file is hosted here as well as in the base repo. **Licence.** PrunaAI publishes PrunaVAED under the LTX-2 licence (`license_name: ltx-2`, base model `diffusers/LTX-2.3-Diffusers`). This converted copy is redistributed under the same LTX-2 Community License Agreement that governs every other file in this repo, with this section as the modification notice (Agreement §3.3). Mirroring and layout conversion do not alter, relax or re-license anything. **Integrity.** `vae_decoder_pruna.safetensors` — 690,025,680 bytes — SHA-256 `891ec5f0680acea9acbdd651c9756a26c5d3298480224356cf5ea86071846f79`.