A modular web UI for AI image, video and audio generation, built on Stable Diffusion, Flux, Wan and other models, with a ComfyUI workflow tab.
TizenryA/colab-gpu-studio
A phone-friendly personal AI studio that runs ComfyUI workflows on on-demand Google Colab GPUs, with Drive caching and task queuing.
- Stars
- 152
- Forks
- 24
- Open issues
- 1
- Last commit
- 9 hours ago
What is it?
Colab GPU Studio is a self-hosted web app for generating images and short videos with ComfyUI from a mobile browser. The React front end is touch-oriented and renders forms from each workflow, while a FastAPI backend, a job queue worker and a timeout watchdog run on your own VPS.
The backend only requests a Google Colab GPU when you start a generation. Model weights and the ComfyUI environment are cached in your Google Drive, so setup work runs on a cheaper CPU runtime and the GPU is released as soon as the output is saved. Queued jobs keep running after the browser is closed.
The project ships with four built-in workflow templates covering dual-sampler text-to-image, single-image text-to-image, first/last-frame image-to-video and multi-reference video generation. Real GPU use is disabled by default, and workflows stay as drafts until you connect your own Google OAuth and enable them.
Key features
- 01Phone-optimized React interface that auto-generates forms from each workflow.
- 02Requests Colab GPUs only when a generation starts and releases them after saving output.
- 03Caches ComfyUI environments and model weights in Google Drive.
- 04Background job queue with a watchdog that cleans up timed-out runs.
- 05Ships with four built-in workflows for image and video generation.
- 06Docker Compose deployment with API, worker and watchdog containers.
Use cases
- Generating images from a phone without managing ComfyUI node graphs by hand.
- Running image-to-video jobs on a rented Colab GPU while keeping the web app on a small VPS.
- Keeping large model downloads cached in Drive so repeated runs skip reinstallation.
- Running a private, single-user AI creation station rather than a multi-tenant service.
Quick start
- 1
Generate your private .env configuration file
uv run python scripts/init_env.py --site-url https://你的域名 - 2
Start the services (api, worker and watchdog containers) with Docker Compose
docker compose up -d --build
Things to know
- Licensed under AGPL-3.0, so modified versions offered over a network must also be published as source.
- Real Google Colab GPU use is disabled by default and needs your own Google OAuth setup, budget limits and workflow review.
- Designed for personal self-hosted use only, with no multi-tenant isolation.
- Built-in real workflows stay as drafts until you verify them, and the README says the offline acceptance CI has not run on GitHub Actions yet.
In collections
Topics
FAQ
- What is colab-gpu-studio?
- A phone-friendly personal AI studio that runs ComfyUI workflows on on-demand Google Colab GPUs, with Drive caching and task queuing. Colab GPU Studio is a self-hosted web app for generating images and short videos with ComfyUI from a mobile browser. The React front end is touch-oriented and renders forms from each workflow, while a FastAPI backend, a job queue worker and a timeout watchdog run on your own VPS.
- Is colab-gpu-studio free to use? What license does it have?
- colab-gpu-studio is open source under the AGPL-3.0 license. Check the license terms before using it in commercial products.
- Is colab-gpu-studio still maintained?
- Actively maintained. The last commit to colab-gpu-studio was on Nov 7, 2026.
- What can I use colab-gpu-studio for?
- Generating images from a phone without managing ComfyUI node graphs by hand. Running image-to-video jobs on a rented Colab GPU while keeping the web app on a small VPS. Keeping large model downloads cached in Drive so repeated runs skip reinstallation. Running a private, single-user AI creation station rather than a multi-tenant service.
- How do I install colab-gpu-studio?
- Generate your private .env configuration file: uv run python scripts/init_env.py --site-url https://你的域名 Start the services (api, worker and watchdog containers) with Docker Compose: docker compose up -d --build
- Where can I get colab-gpu-studio?
- The source code and installation instructions are on GitHub: https://github.com/TizenryA/colab-gpu-studio
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A self-hosted web studio that drives ComfyUI as a Python library to generate audio-video, pictures, 3D models, music and speech on a single NVIDIA GPU machine.
Open-source app for building AI character reference sheets and using those characters in video via character swaps and motion transfer on ComfyUI.
A modular web UI for Stable Diffusion image generation, built to make advanced tools easy to reach and extensible.
A Gradio-based web UI for running Stable Diffusion and newer image/video models, continuing the Forge fork with added model support and optimizations.
Desktop AI art engine that combines a node graph, specialized editors and a 3D director stage for generating short-video, game and advertising assets, with local-first model access.