CPU3DAI tools for 3D, video and audio

MACaQ: Video Generator — What It Does and What You Need to Run It

MACaQ is an open source video generator from the HUAWEI Computing Systems Lab. The tool is built around quantizing models into the GGUF format and is aimed at running generative models of various types locally: text, image, video and audio.

What it does

As described by the author, MACaQ quantizes models without being tied to a specific architecture and supports running models through several backends: llama.cpp, vLLM, Diffusers and PyTorch. This makes it possible to work with text, image, video and audio models in local inference mode.

The repository tags include image generation, video generation, audio generation, work with large language models and GGUF quantization. Support for Ascend hardware is also mentioned.

What you need to run it

The code is written in C++. The authors recommend Python 3.12 and a separate environment for each backend — this requirement is given as described by the developer, not based on measurement results.

The code is distributed under the Apache-2.0 license. It permits use, modification and distribution, including commercial use, provided the authorship notice and license text are retained.

The repository was created on November 10, 2026, and the last code change was on November 12, 2026.

Who it suits

The tool is intended for those who work with running generative models locally and want to use quantization to lower resource requirements. Support for several backends makes it interesting for users who already work with llama.cpp, vLLM, Diffusers or PyTorch and want a single tool for different types of models.

For those looking for a ready-made application with a graphical interface or a cloud service, MACaQ is probably not a good fit: it is a code- and command-line-level tool that requires setting up an environment and understanding how to work with backends.

MACaQ covers the task of quantizing and running models of different modalities locally in a single project. It will suit those who need an open source tool with support for several popular backends and Ascend hardware.

MACaQ pipeline — video generator Quantization and local model launch Model text, image, video, audio GGUF, PyTorch Diffusers Quantization model-agnostic GGUF quantization lower precision no quality loss Backends llama.cpp vLLM Diffusers PyTorch Local launch on-device inference no cloud Ascend, local AI Result generated video local output Export formats: GGUF PyTorch Diffusers License: Apache-2.0 · Open source · HUAWEI Computing Systems Lab
How the MACaQ pipeline works. The diagram is drawn from the tool’s fact sheet.

Fact sheet

Code licenseApache-2.0 source
Python3.12 per the author’s description source
LanguageC++ source
Last code change2026-11-12 source
Repository created2026-11-10 source
Changes often — as of 2026-11-16
GitHub stars319 source
Forks21 source

Values are collected automatically from official sources and were checked on 2026-11-16. Each one links to its source, and values taken from the developer’s pages also carry a verbatim quote — hover over the note. Pricing and versions are shown as of the check date and change most often; verify on the vendor’s site before buying.

See also