Overview
PotionUI is a self-hosted generation studio for images, video, and music on a user's own GPU. It is free and open source and supports Linux and Windows. The project runs multiple model families, including Krea-2, Anima, SDXL, Flux, Wan, LTX, Qwen-Image, MiniMax, and YuE2, on one box. Users pick a model, type, and watch it render live.
It is built for more than one person on the same box. An admin panel supports users and groups, per-user or group presets, model access, LLM configurations, and backends. A configured language model provides an assistant beside generation that can brainstorm and rewrite prompts, edit phrasebook values, and adjust form state, with approval required for actions that change something. PotionUI also mints per-user Model Context Protocol tokens so external MCP clients can drive an instance.
The generate workspace keeps each tab as its own sandbox, and generated media is saved automatically with the exact parameters that produced it. Other features include tags and collections, a Video Director for composing shots in sections, prompt tooling, automations, and plugins. PotionUI is in alpha, so rough edges and breaking changes between releases are expected.
Pricing
Free
Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.
Visit WebsitePrices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.
Use Cases
- Teams use PotionUI to run multiple image, video, and audio models on one GPU box.
- Teams use PotionUI to give multiple users their own logins, presets, and limits on a shared GPU.
- Teams use PotionUI to let an AI assistant brainstorm and rewrite prompts with approval.
- Teams use PotionUI to drive generation from MCP clients such as Claude Desktop.
- Teams use PotionUI to compose video shots in global, timed, and chained sections.
- Teams use PotionUI to automatically save generations with the exact parameters that produced them.
- Teams use PotionUI to extend subsystems through plugins, including alternate inference backends.
- Teams use PotionUI to run on Linux and Windows with their own GPU.
Features
Self hosted generation studio
Image video and music generation
Runs models on your own GPU
Multiple generation models in one interface
Separate user logins for sharing
Per user presets and limits
Live rendering while generating
Linux and Windows support
Reviews
No reviews yet. Be the first to share your experience!