berth
by T-Crypt
Berth is a small programme that makes AI engines, image tools, voice tools and games take turns on a single graphics card, so they stop running out of memory and crashing each other. It also shows which programme is holding the card and who is waiting.
- Stars
- 231 +15 a day, est.
- Licence
- Apache-2.0
- Runs on
- Your computer, with a GPU
- Setup
- Moderate
- Language
- Go
- Age
- 15 days
What you can do with it
- Run a chat model and ComfyUI on the same card without out-of-memory crashes
- Free the card automatically while you play a game, then bring your AI engines back afterwards
- Use one OpenAI-compatible address that starts the right engine for the model you name
Make money with it
Someone could set up and maintain stable local AI workstations for small firms that want private chat and image tools on one card.
An idea, not financial advice. Check the licence before charging for anything built on it.
Ways a business could use it
- Run a local chat model and an image generator on one workstation, avoiding the cost of a second card or cloud GPU hire
- Keep a shared in-house AI machine stable so staff can use chat, image and voice tools without crashes
Can you use it commercially?
Free to use, change and sell, including in paid products. Keep the copyright notice.
Not legal advice. Read the project's licence file before building a product on it.
Getting it running
It is a single Go binary downloaded from the releases page, which you start as a background daemon and then use to launch each engine from the command line.
Follow the official instructions in the project README.
Things to check
- Alpha software (v0.3.6), so expect rough edges and missing features
- The gateway is only a preview
- Mainly aimed at NVIDIA cards
- Windows support is tested by only a few users
Checked 27 Oct 2026. Summary drafted by AI from the project's README.
