Docs

Self-hosting

Run OpenThena — and optionally a LiteLLM proxy — yourself.

OpenThena is a standard Next.js app. Build it and run it anywhere Node runs.

bash

npm install
npm run build
npm run start   # serves on http://localhost:3000

Optional: a real LiteLLM proxy

To centralize keys, budgets, and many providers, run the LiteLLM proxy and point OpenThena at it. A minimal litellm.config.yaml ships in the repo:

yaml

model_list:
  - model_name: mimo-v2.5-pro
    litellm_params:
      model: openai/mimo-v2.5-pro
      api_base: https://api.xiaomimimo.com/v1
      api_key: os.environ/MIMO_API_KEY

bash

pip install "litellm[proxy]"
litellm --config litellm.config.yaml   # serves on :4000

Then tell OpenThena to use it:

bash

LITELLM_BASE_URL=http://localhost:4000/v1
LITELLM_API_KEY=sk-your-litellm-master-key

When LITELLM_BASE_URL is set, all model traffic flows through your proxy — OpenThena becomes a thin client.