TripoSR
github.com
- Free plan
- Linux
- Web
- Self-hosted
What TripoSR is
TripoSR is an open-source model that reconstructs a 3D object from a single image. Developed collaboratively by Tripo AI and Stability AI, it is intended for researchers, developers, and creatives working in 3D generative AI and content creation. The project provides a command-line inference script and a local Gradio app, which can be launched with `python gradio_app.py`. The app includes background removal, foreground-ratio and marching-cubes-resolution controls, and OBJ and GLB outputs. The inference script accepts one or more image paths and can produce a texture instead of vertex colors, with a configurable texture resolution. The project reports generation in less than 0.5 seconds on an NVIDIA A100 GPU; default inference options use about 6 GB of VRAM for one image. Local setup lists Python 3.8 or later, PyTorch, and dependencies from `requirements.txt`; CUDA is optional when available. The code, pretrained models, and interactive demo are released under the MIT license. The listed platforms are Linux, self-hosted, and web.
Who it suits
TripoSR is aimed at researchers, developers, and creatives working on 3D generative AI or 3D content creation. It suits users who want to reconstruct objects from images and can work with its local setup requirements.
Good to know, and what to watch
- Reconstructs 3D objects from a single image
- Includes command-line inference and a local Gradio app
- Supports OBJ and GLB outputs
- Can bake configurable textures
- Code, pretrained models, and demo use the MIT license
- Default inference uses about 6 GB of VRAM per image
- Requires Python 3.8 or later and PyTorch for local setup
- Does not support text-to-3D
Plans
Every plan and priceTripoSR
Free plan
The verdict
TripoSR provides a local image-to-3D workflow with texture options and a browser-based Gradio interface. Its hardware and software requirements are important to check before setting it up.
What the maker publishes
- Demo features
- The Gradio app offers background removal, foreground-ratio and marching-cubes-resolution controls, and OBJ and GLB outputs. source
- Hardware requirement
- Default inference options use about 6 GB of VRAM for a single image input. source
- How it works
- The model generates a 3D model in less than 0.5 seconds on an NVIDIA A100 GPU, according to its README. source
- Input
- The inference script accepts one or more image paths. source
- Intended users
- The project describes its intended audience as researchers, developers, and creatives working in 3D generative AI and 3D content creation. source
- License
- The codebase, pretrained models, and interactive demo are released under the MIT license. source
- License
- The source code and pretrained models are released under the MIT license. source
- License holders
- The MIT license file names Tripo AI and Stability AI in its 2024 copyright notice. source
- Local installation
- The README lists Python 3.8 or later, PyTorch, and dependencies installed from requirements.txt; CUDA is optional if available. source
- Local interface
- The repository includes a Gradio app that can be started with python gradio_app.py. source
- Memory requirement
- The default inference options use about 6 GB of VRAM for one image input. source
- Output options
- The inference script can bake a texture instead of using vertex colors and lets users specify texture resolution. source
- Pricing
- The TripoSR repository describes the model as open source under MIT, but does not state a subscription price for TripoSR. source
- Product identity
- TripoSR is an open-source model for fast 3D reconstruction from a single image, developed collaboratively by Tripo AI and Stability AI. source
- Requirements
- The setup instructions list Python 3.8 or newer and recommend installing CUDA if available. source
- Run locally
- The repository provides a command-line inference script and a local Gradio app. source
- Single-image input
- TripoSR reconstructs 3D objects from a single image. source
- Speed
- The project says it can generate a 3D model in less than 0.5 seconds on an NVIDIA A100 GPU. source
- Texture output
- Inference can output a texture instead of vertex colors using the --bake-texture option, with texture resolution configurable. source
- What it does
- TripoSR reconstructs a 3D object from a single image using an open-source model. source
Checked against the maker's pages on 8 Oct 2026.