Downloads · 30 days
0
Trellis_2_RTX3060_2.10.0_cu130
ayushroutray5777/Trellis_2_RTX3060_2.10.0_cu130
Trellis_2_RTX3060_2.10.0_cu130 is a machine learning model from ayushroutray5777. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
[](https://huggingface.co/PixWizardry/AITrellis2-WHLs-RTX-PRO-6000/tree/main) [](https://www.python.org/) [](https://developer.nvidia.com/cuda-toolkit)
Access
Public
Updated May 12, 2026
Repo size
6.1 MB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.whl5.9 MB · 100%
At a glance
- License
- mit
- Access
- Public
- Created
- May 12, 2026
- Updated
- May 12, 2026
- SHA
- 5d3478ee
Model card
From the Hugging Face model README
⚡ Custom Wheels for RTX 3060 OR 3XXX (Ampere) - ComfyUI Trellis2
TRELLIS 2 Windows CUDA 13 Wheels (Torch 2.10)
Prebuilt and locally rebuilt Windows wheels for running:
- TRELLIS.2
- ComfyUI-Trellis2
- ComfyUI-Trellis2-GGUF
- comfyui-if_trellis
on modern Windows CUDA environments.
These wheels were rebuilt from source specifically for:
- Python 3.12
- PyTorch 2.10.0 + CUDA 13.0
- Windows 11
- RTX 3060 (SM86 tested)
Why This Repository Exists
Most prebuilt TRELLIS/Torch CUDA wheels currently target:
- Torch 2.8 / 2.9
- CUDA 12.x
- Python 3.10 / 3.11
Using them on newer environments often causes:
- DLL load failures
_C.pydcrashes_nvdiffrast_cimport errors- CUDA ABI mismatches
- Torch symbol resolution failures
- Windows fatal exception
0xc0000139
This repository contains rebuilt and tested wheels that successfully work together on:
Python 3.12.11
Torch 2.10.0+cu130
CUDA Toolkit 13.0
Windows 11
RTX 3060
Included Packages
| Package | Status |
|---|---|
| kaolin | rebuilt |
| nvdiffrast | rebuilt |
| cumesh | rebuilt |
| flex_gemm | rebuilt |
| o_voxel | rebuilt |
Tested Stack
| Component | Version |
|---|---|
| OS | Windows 11 |
| Python | 3.12.11 |
| PyTorch | 2.10.0+cu130 |
| CUDA Toolkit | 13.0 |
| GPU | NVIDIA RTX 3060 |
| Visual Studio | VS2022 Build Tools |
| Generator | Ninja |
Installation
1. Install PyTorch
Install the exact Torch version first:
pip install torch==2.10.0+cu130 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu130
2. Install Wheels
Install in this exact order:
pip install kaolin-0.18.0-cp312-cp312-win_amd64.whl
pip install nvdiffrast-0.4.0-cp312-cp312-win_amd64.whl
pip install cumesh-0.0.1-cp312-cp312-win_amd64.whl
pip install flex_gemm-1.0.0-cp312-cp312-win_amd64.whl
pip install o_voxel-0.0.1-cp312-cp312-win_amd64.whl
Verification
Run:
python -c "import torch; import kaolin; import nvdiffrast.torch as dr; import cumesh; import flex_gemm; import o_voxel; print('ALL TRELLIS EXTENSIONS OK')"
Expected output:
ALL TRELLIS EXTENSIONS OK
Important Notes
These wheels are ABI-specific.
They are intended ONLY for:
Python 3.12
Torch 2.10.0+cu130
CUDA 13.0
Windows x64
Using different:
- Torch versions
- CUDA versions
- Python versions
may cause:
- DLL failures
- unresolved symbols
- extension crashes
Build Notes
Several packages required source patching for:
- Torch 2.10 compatibility
- CUDA 13 compatibility
- MSVC 2022 compatibility
- Eigen template parsing
- Windows namespace conflicts
- CUDA extension rebuilding
Recommended Usage
These wheels are intended for:
- TRELLIS.2
- ComfyUI-Trellis2
- ComfyUI-Trellis2-GGUF
- comfyui-if_trellis
- Windows sparse CUDA mesh workflows
Credits
Original projects:
- NVIDIA Kaolin
- NVIDIA nvdiffrast
- Microsoft TRELLIS.2
- JeffreyXiang/CuMesh
- JeffreyXiang/FlexGEMM
Windows rebuilds and compatibility fixes by:
Ayush Routray
- License
- mit
- Created
- May 12, 2026
- Updated
- May 12, 2026