LA

LatentSync 1.6

LatentsyncGGUF by audio-cppUpdated 2026-10-21
Download on Hugging Face
Parameters1.37B
Context—
Layers—
Architectureaudiocpp
Smallest quant2.1 GB
Quick answer

LatentSync 1.6 needs about 3.6 GB of memory at F16 (a 2.7 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 3 quants, from 2.1 GB (Q8_0) to 5.4 GB (F32).

LatentSync 1.6 – audio‑driven lip sync for 512×512 videos

LatentSync 1.6 is a 1.37 billion‑parameter audio‑to‑video model that synchronizes lip motion in existing videos with supplied speech. It builds on the 1.5 architecture and was trained on 512×512 resolution videos to improve detail. The model is available in GGUF format for audio.cpp inference.

Good at

  • High‑resolution lip sync
  • Fast inference with audio.cpp
  • Compatible with 1.5 checkpoints

Limits

  • No vision support beyond lip sync
  • Requires InsightFace face detection weights
  • License includes non‑commercial research restriction

Best for

  • Video dubbing
  • Lip‑sync correction in post‑production
  • Research on audio‑visual alignment

Who it’s for. Developers and researchers needing precise lip‑sync for 512×512 videos. The model is suitable for projects that can accommodate the non‑commercial license terms and the need for InsightFace weights.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q8_02.1 GB 2.8 GB
Fits Get file →
F162.7 GB 3.6 GB
Fits Get file →
F325.4 GB 6.6 GB
Fits Get file →

For 8 GB VRAM, use the Q8_0 quant (2.1 GB file, 2.8 GB memory at 8k context). If more VRAM is available, F16 (2.73 GB file, 3.6 GB memory) offers higher precision. F32 (5.45 GB file, 6.6 GB memory) is only recommended for systems with large memory budgets.

What GPU do you need for LatentSync 1.6?

These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: the best quant that fits is F16 (2.7 GB file, about 3.6 GB in use).
  • 6 GB VRAM: the best quant that fits is F16 (2.7 GB file, about 3.6 GB in use).
  • 8 GB VRAM: the best quant that fits is F32 (5.4 GB file, about 6.6 GB in use).

Which LatentSync 1.6 quant should you choose?

F16 (2.7 GB) is the middle option in this repository. The unquantized model. It gives original quality and needs the most memory, so it is mainly for conversion or reference.

If memory is tight, the smallest file is Q8_0 at 2.1 GB, with the quality trade-offs that come with fewer bits. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureaudiocpp
Parameters1.37B

Source

Quantized byaudio-cpp
Licenseother
Downloads (HF)16,650
Last updated2026-10-21

Frequently asked questions about LatentSync 1.6

How much VRAM does LatentSync 1.6 need?

About 3.6 GB at F16 with an 8K context window. That is the 2.7 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run LatentSync 1.6 on an 8 GB GPU?

Yes, with F32: the 5.4 GB file needs about 6.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of LatentSync 1.6 GGUF?

It depends on the quant: Q8_0 is 2.1 GB, F16 is 2.7 GB, F32 is 5.4 GB.

Where can I download LatentSync 1.6 GGUF?

The files are on Hugging Face in the audio-cpp/LatentSync-1.6-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does LatentSync 1.6 use?

Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run LatentSync 1.6 on 8 GB VRAM?

Yes, the Q8_0 quant requires 2.1 GB file size and estimates 2.8 GB memory at 8k context, fitting within 8 GB VRAM.

Does the model support other languages?

The model does not specify language support; it focuses on audio‑driven lip sync for 512×512 videos.

What are the licensing restrictions?

The model is under an other license, but includes InsightFace weights restricted to non‑commercial research. Review upstream licenses before redistribution.

Similar models

Browse all →

More from audio-cpp

All by audio-cpp →

Other tiny models

All tiny models →