#109109th of 126 in ML frameworks and librariesFor image, video and design, 3d
gsplat
CUDA library for rasterizing 3D Gaussians with PyTorch bindings
- Stars
- 5.9k
- License
- Apache-2.0
- Last commit
- Sep 2026
- Last release
- Jul 2025
- Language
- Python
4,785 on 10 Apr 20265,854 on 30 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
gsplat is a Python library with CUDA kernels for differentiable rasterization of 3D Gaussians, based on the 3D Gaussian Splatting paper. It includes training examples for COLMAP and NCore v4 captures, 3DGUT support, MCMC densification, LiDAR and fisheye camera models, and an experimental fp16 inference renderer. The README reports up to 4x less GPU memory and 15% less training time than the official implementation.
Who it is for: Researchers and engineers training or rendering Gaussian splats on NVIDIA GPUs
Strengths
- Reports up to 4x less GPU memory and 15% less time than the official implementation
- Supports 3DGUT, MCMC densification, batching, and multiple camera and LiDAR sensor models
- Pre-compiled wheels for Linux and Windows on selected Python, torch and CUDA combinations
- Apache-2.0 with a JMLR paper, docs site and benchmark scripts
Weaknesses
- Requires CUDA; PyPI install compiles kernels on first run via JIT
- v1.6.0 features are on main only and not yet on PyPI
- Library, not an app: no GUI for capture-to-splat beyond example scripts
- Last tagged release was July 2025; main will require PyTorch 2.7+
What it needs
- GPU required
- Needs PyTorch, CUDA, ninja, numpy, jaxtyping, rich
- Models: 3D Gaussian Splatting, 3DGUT, COLMAP datasets, NCore v4
Quick answers
Is gsplat open source?
Yes. Its code is on GitHub under the Apache-2.0 license.
Does gsplat need a GPU?
Yes, it needs a GPU.
Also in ML frameworks and libraries
See all 126| Rank | Project | Score |
|---|---|---|
| 1 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 95 out of 100 |
| 2 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 94 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 93 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 93 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 91 out of 100 |