https://github.com/ayan4m1/vast-ai-provisioner
Build a template on the fly for deployment to Vast.ai
https://github.com/ayan4m1/vast-ai-provisioner
cli vast-ai
Last synced: about 1 month ago
JSON representation
Build a template on the fly for deployment to Vast.ai
- Host: GitHub
- URL: https://github.com/ayan4m1/vast-ai-provisioner
- Owner: ayan4m1
- License: mit
- Created: 2025-07-03T16:44:25.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-08-26T11:51:09.000Z (11 months ago)
- Last Synced: 2025-11-11T14:33:51.683Z (9 months ago)
- Topics: cli, vast-ai
- Language: Shell
- Homepage:
- Size: 175 KB
- Stars: 2
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# vast-ai-provisioner
This script will provision a [Vast.ai](https://vast.ai) instance with a specific template which can be overridden and customized beyond what is available with Vast's template system. The repo currently provides the following templates:
- [ComfyUI](https://github.com/comfyanonymous/ComfyUI) - A node-based AI image/video generation suite
- [FP-Studio](https://github.com/FP-Studio/FramePack-Studio) - A FramePack-based video editor
Additionally, there are several modules available to add to the base ComfyUI install:
- Animagine: Animagine XL 4.0, a best-in-class anime model
- Flux Dev: Adds Flux 1.Dev and ControlNet
- Flux Kontext: Adds Flux Kontext for image editing
- Pixel Art: Adds Pixel Art LoRAs for SDXL
- SDXL: Adds SDXL and ControlNet
- Upscalers: A collection of high quality 2x and 4x upscalers
- Utility Nodes: A set of Comfy nodes to make building workflows easier
- WAN2.2: A powerful text-to-video or image-to-video model
## Requirements
- Node >=22
## Usage
```sh
npm install -g vast-ai-provisioner
vast-ai-provisioner -h
OR
npx vast-ai-provisioner -h
```
## Development
```sh
corepack enable
yarn install
yarn run build
node lib/index.js
```
Template is an optional argument - if you do not provide it, the application will ask you which one you want to deploy.